QUALITY

CONTROL OF
DYNAMIC X-RAY
IMAGING
SYSTEMS
EFOMP PROTOCOL
VERSION 01. 2024

[Link]
version 01.2024 1
This page intentionally left blank.

2 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Preface

This document was developed by an EFOMP Working Group with the aim of providing guidance on Accept-
ance, Commissioning and Constancy performance testing over the life of an x-ray imaging system. The main
motivation for this document is to provide Medical Physics Experts (MPE) with unified guidance across
Europe on the assessment of dynamic x-ray imaging systems, in an area that currently lacks such advice.
Current testing practice is quite disparate, with specific and varied performance criteria being applied from
country to country. It can be difficult for manufacturers to meet all of these different requirements and the
introduction of a standardised protocol should simplify the situation, ensuring a consistent and compre-
hensive set of tests.

Image quality assessment of these systems has always been a challenge, given the dynamic nature of the
images and the fact that a visual evaluation often had to be performed ‘on the spot’ by the MPE. There have
been a number of developments in the area of image quality assessment using computational methods and
a further aim was to provide information on this topic.

Scope

This document provides guidance for testing dynamic x-ray systems. We will use the term ‘dynamic x-ray
imaging system’ as a blanket term, covering simple mobile fluoroscopy to complex angiography devices.
Not covered is the testing of cone beam computed tomography (CBCT) modes also referred to as 3D
rotational angiography (3DRA) imaging. In these modes, the x-ray tube/detector assembly is rotated by
at least 200° around the patient as a number of low dose projection images are acquired. These are then
reconstructed to generate a volumetric dataset. Clinical applications include planning in stereotactic radio-
surgery and angiography examinations such as imaging of aneurysms in neuroradiology and the evaluation
of stent placement (Doerfler et al., 2015; Fahrig et al., 2021). Guidance on testing these systems is available
in a separate EFOMP protocol (de las Heras Gala et al., 2017).

version 01.2024 3
Working Group

Members EFOMP Scientific Committee chairpersons

Annalisa Trianni (Italy) Yolanda Prezado (Spain)

Andy Rogers (UK) Brendan McClean (Ireland)
Marco Bertolini (Italy) Eeva Boman (Finland)
Hilde Bosmans (Belgium)
Nicholas Marshall (Belgium)
Nicoletta Paruccini (Italy)
Michael Sandborg (Sweden)
Alexander Schegerer (Switzerland)
Lucie Sukupova (Czech Republic)
Diego Trevisan (Italy)

4 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Consultants: Observers:

Steve Balter (US) Carri Borras (US)

Jérémie Dabin (Belgium) Richard Elek (Hungary)
Lynn Gaynor (Ireland) Pei-Jan Paul Lin (US)
Jan Jans (The Netherlands) Desislava Kostova-Lefterova (Bulgaria)
Shahed Khan (UK) Ivan Lasic (Bosnia and Herzegovina)
Markus Lendl (Germany) Sabina Strocchi (Italy)
Francoise Malchair (Belgium) Giovanna Venturi (Italy)
Despina Papadopoulou (Greece) Raffaelle Villa (Italy)
Kevin Wunderle (US)

version 01.2024 5
TABLE OF CONTENTS
SECTION A – GENERAL BACKGROUND 10
A.1. DYNAMIC X-RAY IMAGING SYSTEMS 10
A.2. TESTS AND CRITERIA 10
A.2.1 Tests 10
A.2.2 Criteria 12
A.3. MEDICAL PHYSICS EXPERT ROLE AND RESPONSIBILITIES 12
A.4. INSTRUMENTS 13
A.5. PROTOCOL OVERVIEW 13
A.5.1 Mechanical and geometrical parameters 15
A.5.2 X-ray tube and generator 15
A.5.3 Automatic Exposure Control (AEC) 15
A.5.4 Dose indicators 15
A.5.5 Skin Dose Map 16
A.5.6 Detector 17
A.5.7 Image Quality 19
A.6. SAFETY 28
A.7. OTHER GUIDANCE 29

SECTION B – PROTOCOL 30
B.1. MECHANICAL AND GEOMETRICAL PARAMETERS 30
B.1.1 Determination of source location and minimum source – skin distance 30
B.1.2 Minimum field size 30
B.1.3 Beam alignment 31
B.1.4 Correspondence between X-ray field and effective image receptor size 32
B.1.5 Verification of displayed distances 33
B.2. X-RAY TUBE AND GENERATOR 33
B.2.1 Tube voltage accuracy 33
B.2.2 Minimum HVL/filtration evaluation 34
B.2.3 Half value layer (HVL) evaluation for clinical spectra 35
B.2.4 Normalised air kerma 36
B.2.5 Leakage radiation 37
B.3. AUTOMATIC EXPOSURE CONTROL 37
B.3.1 AEC function 37
B.3.2 Comprehensive evaluation of AEC 39
B.4. DOSE INDICATORS 40
B.4.1 Air-kerma – Area product and rate display accuracy 40
B.4.2 Manufacturer specified Air-kerma rates 42
B.4.3 Measurement of table and mattress attenuation 42
B.4.4 Limiting Air Kerma Rate 43
B.5. SKIN DOSE MAPS 44
B.5.1 Skin dose map – Acceptance 44
B.5.2 Skin dose map – Commissioning 44
B.6. DETECTOR 45
B.6.1 Response Function 45
B.6.2 Detector presampling Modulation Transfer Function (MTF) 47

6 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 B.6.3 Image detector Brightness Non Uniformity (BNU) 48
B.6.4 Variance and SNR Image 49
B.6.5 Noise Decomposition using Variance 49
B.6.6 Noise Power Spectrum (NPS) 50
B.7. IMAGE QUALITY 51
B.7.1 Signal-to-noise ratio rate, SNR 2rate 51
B.7.2 Threshold-contrast detail detectability (TCDD) – Statistical method 52
B.7.3 Threshold-contrast detail detectability (TCDD) – CHO readout 54
B.7.4 Threshold-contrast detail detectability (TCDD) – Human visual scoring 55
B.7.5 Limiting spatial resolution (LSR) 56

SECTION C – REFERENCES 57

SECTION D – APPENDICES 63
D.1. Appendix 1. Radiation Safety Procedures 63
D.2. Appendix 2. Phantoms 65

version 01.2024 7
LIST OF ABBREVIATIONS

AAPM American Association of Physicist in Medicine

AEC Automatic Exposure Control
AIFM Associazione Italiana di Fisica Medica - Italian Association of Physicist in Medicine
AP AnteroPosterior
BNU Brightness Uniformity
BSS European Basic Safety Standards
CBCT Cone Beam Computed Tomography
CDDISC Contrast-Detail phantom with cylindrical holes
CDRAD Contrast-Detail phantom for Radiography
CTSM Threshold Contrast for the Statistical Method
CTPV Threshold Contrast for the Statistical Method in terms of Pixel Values
CHO Channelized Hotelling Observer
CT Computed Tomography
CV Coefficient of Variation
DMS Dose Management System
DQE Detective Quantum Efficiency
DICOM Digital Imaging and Communication in Medicine
EFOMP European Federation of Organizations For Medical Physics
ESAKR Entrance Surface Air Kerma Rate
EU European Union
FN False Negative
FNF False Negative Fraction
FoV Field Of View
FP False Positive
FPF False Positive Fraction
FPD Flat Panel Detector
HVL Half Value Layer
IAEA International Atomic Energy Agency
IBSS International Basic Safety Standards
ICRU International Commission on Radiation Units and measurements
IEC International Electrotechnical Commission
IPEM Institute of Physics and Engineering in Medicine
IR-IAK Image Receptor Incident Air Kerma
KAP Air Kerma Area Product
Ka,r Reference Air Kerma
LCDD Low-Contrast Detail Detectability
LSR Limiting Spatial Resolution
MAFC Multiple Alternative Forced Choice
MO Model Observer
MPE Medical Physics Expert
MPV Mean Pixel Value
MTF Modulation Transfer Function
NEMA National Electrical Manufacturers Association
NPS Noise Power Spectrum
PERP Patient Entrance Reference Point
PKA Air Kerma Area Product

8 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

P(t)KA Air Kerma Area Product Rate
PSD Peak Skin Dose
PMMA Polymethyl Methacrylate
PTB Physikalisch Technische Bundesanstalt - German National Metrology Institute
PV Pixel Value
QA Quality Assurance
QC Quality Control
RDSR Radiation Dose Structured Report
ROI Region Of Interest
ROC Receiver Operating Characteristic
SDM Skin Dose Mapping
SDNR Signal Difference-to-Noise Ratio
SI International System of Units
SID Source to Image Detector Distance
SM Statistical Method
SOD Source to Object Distance
SNR Signal-to-Noise Ratio
TCD Threshold Contrast Detection
TCDD Threshold Contrast Detail Detectability
TIQ Technical Image Quality
TN True Negative
TNF True Negative Fraction
TO Test Object
TP True Positive
TPF True Positive Fraction
VERIDIC Validation and Estimation of Radiation Skin Dose in Interventional Cardiology
VGC Visual Grading Characteristics
VGR Visual Grading Regression
WHO World Health Organization
XRII X-Ray Image Intensifier

version 01.2024 9
SECTION A - GENERAL BACKGROUND

A.1. DYNAMIC X-RAY IMAGING SYSTEMS

The dynamic x-ray imaging systems covered by this protocol generate a temporal sequence of images that
may or may not be stored by the system. The term ‘dynamic x-ray imaging system’ covers a wide range of
devices, and these can be broadly grouped as follows:

1. Complex C-arm systems used for the purpose of diagnosis and treatment/intervention. A broad range of ap-
plications is covered including abdominal radiology, cardiology, neuroradiology and peripheral angiography
2. Bi-plane devices, generally used for cardiac and neuroradiology examinations.
3. Radiography/fluoroscopy systems, typically with an overcoach x-ray tube and tilting table, generally used
for gastrointestinal examinations and contrast examinations in the radiology department
4. Mobile C-arm systems, for example used in operating theatres
5. Mini C-arms, used for the imaging of extremities
6. Radiography x-ray systems, with a fluoroscopy mode designated for patient positioning

The above is not an exclusive list and the protocol should be applied to all dynamic x-ray devices used at the
clinical site.

Two main modes are used in dynamic x-ray systems. The term ‘fluoroscopy’ is applied to the mode where a
sequence of images is generated and used as part of the examination. These images are not generally stored.
Once the fluoroscopy sequence is stopped by the operator then the last frame or weighted combination of
the last few frames is held on the display, as a ‘last image hold’ image. Many systems also allow images from
the most recently performed fluoroscopy sequence to be stored or archived as an image sequence, using a
function such as ‘fluoroscopy store’. In addition to performing fluoroscopy, systems can acquire images that
are stored as ‘acquisition’ images (sometimes called cineangiography). This can be single shot, or at some
frame rate, typically ranging from 1 to 4 frame/s for radiology/angiography applications up to 10 or 15 frame/s
even or higher for cardiac acquisitions.

A.2. TESTS AND CRITERIA

A.2.1 Tests
The testing procedures applied by the MPE vary in scope and depth depending on the type and intention of
the test and the equipment being evaluated. Before testing a system, the main clinical applications should be
established along with the most commonly used fluoroscopy and acquisition modes. These will be the modes
evaluated in the Constancy testing of the device and depend on the Acceptance and Commissioning testing.
Tests should be relevant to clinical practice on the unit and should be able to uncover changes that can
potentially affect the clinical performance of the system. Dose Management Systems (DMS) can also greatly
help in establishing system usage and which fluoroscopy and acquisition modes are commonly used. This
would aid the MPE to ensure the relevance of the tests. Simplified tests can be carried out by technologists/
radiographers at frequencies determined by national guidance or recommended by the MPE.

Definitions of the various test types and how they fit in with the lifecycle of the equipment are largely con-
sistent in the literature although some variations are seen (IPEM, 2005; European Commission, 2012; Dance
et al., 2014; Jones et al., 2014). Note that there is likely to be some variation in test definitions and the person
assigned with the duty of performing tests for the different national protocols already in place; these must be
taken into consideration when applying this protocol. Focus in this protocol is on Acceptance, Commissioning
and Constancy QC tests.

10 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Acceptance Tests
Acceptance tests ensure that new or replacement equipment complies with the specifications made at the
tendering stage and performance meets that specified by the vendor for the system. When verifying aspects
of system performance (including radiation safety assessments), measurements may have to be made using
the methods specified by the equipment manufacturer/vendor. This will normally involve representatives of
the vendor/installer and the MPE. These tests are performed to ensure that the system complies with accept-
ability requirements that may be in place at a regulatory (national) level and are required before the first use
of the system on a patient. Acceptance tests are also required when new or replacement components or the
installation/upgrade of software leads to a clinically important change in performance.

Commissioning Tests
Commissioning tests follow Acceptance testing and clinical protocol development. They characterise the
dose and image quality performance of the system. Clinical protocol development will involve inspection and
possible adjustment of the default clinical protocols, pre-installed on the system, to meet the clinical require-
ments, and will be done in conjunction with the radiological practitioner, x-ray technologists/radiographer,
the MPE and the installer’s applications specialist. This can be an extended process that happens over a few
weeks or months. Commissioning is the first stage in the optimisation process. The performance must be
achieved whilst staying within regulatory limits that are in place.
It can be very helpful to record the values of various fluoroscopy and acquisition settings that influence
dose and image quality delivered by a given imaging mode/protocol. This can help to avoid over testing
by identifying modes that have different names but the same/similar underlying dose and image quality
characteristics. Even if this can only be done to a limited extent, this will help the MPE’s understanding of the
system and improve the relevance of the tests.
A (sub)set of Commissioning tests should be performed after the Acceptance test of newly installed compo-
nents, to establish whether or not new baselines for longitudinal testing are required.
If, subsequent to initial Commissioning, a change is made to an imaging protocol that substantially affects
dose and image quality, then appropriate re-Commissioning should be undertaken and any affected baselines
should be adjusted to reflect the results for the new protocols.

Constancy Tests
These tests have variously been called ‘quality control’ tests or ‘routine’ tests.
They are a set of tests (normally a subset of the full Acceptance and Commissioning tests) performed at reg-
ular intervals, with the aim of confirming that selected performance parameters remain within their respective
baseline values. Any changes in performance should be documented, and any corrective action should be
reported to the department.
At Constancy testing the tester should record the parameter settings for the protocols as defined by the MPE
during Commissioning. These can then be compared to values at previous visits and to those at Commissioning.
After testing, it is important to ensure that the system is left in/returned to its standard clinical configuration.
The x-ray tube port should be clear of any (metal) filters that were used for the tests.

User Quality Control Mode

Tracking or auditing of protocol parameters is made much easier if the User Quality Control Mode, as de-
scribed in NEMA standard (NEMA, 2018), is implemented on the system. This Standard enables the settings of
all imaging protocols and modes to be exported to a spreadsheet for analysis. Changes in parameter settings
are indicated, making the tracking of changes in protocols straightforward. When investigating changes in
system performance, this can be separated into the identification of programmed (intentionally or uninten-
tionally) changes in protocol settings (software) and actual changes in component (hardware) performance.
Implementation of the NEMA User Quality Control Mode has additional advantages and it is hoped that more
systems will include this. Manual control of the x-ray factors is possible, which makes tube and generator

version 01.2024 11
testing and calibration of the kerma-area product (PKA) meter more straightforward and should improve
reproducibility. Manual control of x-ray factors in conjunction with access to ‘For Processing’ images will allow
the explicit assessment of x-ray detector performance. While detailed evaluation of the x-ray detector is not
necessarily done at every routine visit, explicit testing of x-ray tube and detector is very helpful when trying
to isolate or troubleshoot problems reported about the system in general by the operators.

Daily Quality Control

As part of a comprehensive quality assurance system, it is important for the users (normally radiographers/
technologists) to undertake a set of quick tests prior to first use of the day. The rationale for undertaking
these tests, which take only a few minutes, is to ensure that the equipment is working before the first patient is
prepped, i.e., to exclude the possibility of the patient being prepped unnecessarily. A list of suggested tests is:

1. Ensure all powered movements of the gantry work properly (where applicable)
2. Ensure that the table and detector powered movements work properly (where applicable)
3. Ensure that radiation can be emitted, and different fluoroscopy modes/magnifications can be selected
(as applicable)
4. Ensure the collimators (and wedges where applicable) drive in and out
5. Disable x-rays then perform World Health Organisation type checklist tasks (see for example Appendix D.1)

If any of the above tests indicate a unit that is not working as usual, the user should seek advice to ascertain
whether the unit may be used on patients.

A.2.2 Criteria
A test is applied, and the results compared against Pass/Fail Criteria when available. These criteria can take
different forms and are specified in the description of each test.
- Action level: corrective action is required if the value of the test parameter is out of the defined range for
the test.
- Baseline value: a value of a parameter established in a Commissioning test and used for comparison in
constancy testing. Often the percentage change in the value of the parameter from the baseline value is
computed and if a subsequent measurement falls outside of test-specific tolerance levels then some form
of corrective action is required.
- Limiting value: the maximum or minimum value of a parameter considered acceptable by either interna-
tional standards or national regulations.

A.3. MEDICAL PHYSICS EXPERT ROLE AND RESPONSIBILITIES

It is the responsibility of the Medical Physics Expert to ensure that the frequency and depth of testing is
commensurate with the imaging applications undertaken on a particular system.
The goal is not just to apply a set of pass-fail tests and move on to the next system. The MPE should have
some knowledge of the imaging tasks undertaken on the system, ensure that relevant Acceptance and
Commissioning tests have been carried out and that the system is set up to perform the clinical tasks
routinely undertaken on the system. Constancy testing can then take over, ensuring that the system remains
within the baselines set at Commissioning. This should involve a regular audit of the system protocols.
Additionally, the duties of the MPE also include equipment specification and evaluation, which are impor-
tant stages within the Quality Assurance (QA) framework (Dance et al., 2014), however the focus in this
protocol is on Acceptance, Commissioning and Constancy tests.

12 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

A.4. INSTRUMENTS

Calibration
In order to obtain meaningful readings from your instrument you need to understand its response. This can
be achieved through testing the instrument yourself under different circumstances or calibrating it for the
quantities you intend to measure, e.g. air kerma, kVp, HVL etc. The manufacturer of your instrument should
be able to give advice on where you can obtain a calibration certificate if one is not provided at purchase.
The calibration is performed against a reference instrument (e.g., an air kerma detector, high voltage di-
vider etc.), which in turn is traceable to a primary standards laboratory, e.g. PTB, Germany. The expanded
uncertainty for the calibration factor is stated as the standard uncertainty multiplied with a coverage factor,
k=2, which for a normal distribution corresponds to a coverage probability of 95% (true value lies within the
coverage interval 95% of the time). If your instrument has been serviced, it is wise to have it calibrated or at
least compare its reading with another calibrated instrument to make sure it is working properly.

Accuracy and Precision

You also need to check the specification of your instrument with regards to the accuracy of any derived
quantity so that you can draw the correct conclusion with regards to any measured change in the quantity
of interest. For example, your multimeter estimates HVL with an uncertainty (k=2) of ±10% or ±0.2 mm
Al. When applying uncertainties to the limiting or baseline values against which the measurement will be
compared, the MPE may wish to distinguish between performance metrics such as kV and limiting values
set by standards, e.g. maximum air kerma rate. In the latter you may wish to subtract your instrument's
uncertainty from the limiting value to ensure a high probability of conformance.
It is also useful to know the precise location of the instrument’s reference point. The instrument specifica-
tion is also useful as it will mention the range of, for example, air kerma rates when you can expect reliable
readings. Measurement outside this specified range may reduce the precision. Ensure that you understand
your dosimeter’s response to backscattered radiation: add a back-scatter fraction when either adding or
removing such effects from measured values, as appropriate.

Energy dependence
Some instruments may have a significant so-called energy-dependence and therefore the calibration factor
for the reference beam quality, e.g. RQR5 (70 kV, HVL=2.58 mmAl) at the calibration facility, may be differ-
ent from the factor for the beam qualities used in your hospital. This can be an issue with air kerma area ion
chambers (PKA meters) that contain thin metallic conductive layers when you measure PKA for an imaging
system that uses varying filtration and tube voltages. Therefore, check your instruments’ response to dif-
ferent beam qualities and, if possible, use instruments with minimal or at least known energy dependence.
Useful information regarding dosimetry and use of dosimeters is given in various IAEA documents including
IAEA TRS457 (IAEA 2007).

A.5. PROTOCOL OVERVIEW

This protocol describes an extensive array of tests (Table 1), methods and test frequencies that can be
applied by MPE to assess the performance of dynamic x-ray imaging systems used at a medical facility.
Tests are divided into 7 groups covering different aspects of the equipment:
1) Mechanical and geometrical
2) X-ray tube
3) Automatic Exposure Control
4) Dose indicators
5) Skin Dose Map

version 01.2024 13
6) Detector
7) Image quality
In the following paragraphs the rationale and general introduction to the different tests are described
where it is felt that some guidance is required. Not every test described in Section B [the protocol] has a
corresponding description in Section A.

Table 1 Tests and test frequencies reported in section B of the present document. The numbers in the Test column indicate the
subsection of the protocol and the specific test, respectively. X = test to be performed.

Test Acceptance Commissioning Constancy

B.1 MECHANICAL AND GEOMETRICAL PARAMETERS
B.1.1 Determination of source location and minimum source - skin distance X
B.1.2 Minimum field size X
B.1.3 Beam alignment X
B.1.4 Correspondence between X-ray field and effective image receptor size X X
B.1.5 Verification of displayed distances X
B.2 X-RAY TUBE AND GENERATOR
B.2.1 Tube voltage accuracy X X
B.2.2 Minimum HVL/filtration evaluation X
B.2.3 Half value layer (HVL) evaluation for clinical spectra X X
B.2.4 Normalised air kerma and air kerma rate X
B.2.5 Leakage radiation X
B.3 AUTOMATIC EXPOSURE CONTROL
B.3.1 AEC function X
B.3.2 Comprehensive evaluation of the functionality of AEC X X
B.4 DOSE INDICATORS
B.4.1 Air-kerma – Area product and rate display accuracy X X
B.4.2 Manufacturer specified Air-kerma rates X X
B.4.3 Measurement of table and mattress attenuation X
B.4.3 Limiting air kerma rate X X
B.5 SKIN DOSE MAPS
B.5.1 Skin dose map – Acceptance X
B.5.2 Skin dose map – Commissioning X
B.6 DETECTOR
B.6.1 Response Function X
B.6.2 Detector presampling Modulation Transfer Function (MTF) X
B.6.3 Image detector Brightness Non Uniformity (BNU) X X
B.6.4 Variance and SNR Image X
B.6.5 Noise Decomposition using Variance X X
B.6.6 Noise Power Spectrum (NPS) X X
B.7. IMAGE QUALITY
B.7.1 Signal-to-noise ratio rate, SNR2rate X X
B.7.2 Threshold-contrast detail detectability (TCDD) – Statistical method X X
B.7.3 Threshold-contrast detail detectability (TCDD) – CHO method X X
B.7.4 Threshold-contrast detail detectability (TCDD) – Human visual scoring X X
B.7.5 Limiting spatial resolution (LSR) X X

14 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

A.5.1 Mechanical and geometrical parameters
An imaging system requires mechanical stability in order to achieve the intended operational performance. An
assessment of mechanical and geometrical performance assures the MPE that the system is properly set up.

A.5.2 X-ray tube and generator

The x-ray tube and generator are the first components in the production of known and stable X-ray beam.
Therefore, characterization of the performance is a crucial element of the overall quality control programme.
Moreover, tube and generator performance influences patient dose and image quality.

A.5.3 Automatic Exposure Control (AEC)

The AEC system of the imaging equipment ideally maintains image quality or dose based upon the detector
signal and potentially reduces patient skin entrance dose to a minimum regardless of selected protocol
parameters and thickness as well as composition of the part of the patient's body that is being imaged.
The AEC system may automatically control many imaging parameters (e.g., focal spot exposure time, tube
voltage, tube current, filter, or/and pulse width), simultaneously, based upon the detector signal. Some
recent units may adjust the detector dose as a function of attenuator thickness.
Given the multitude of protocols available on modern systems, a systematic method should be used to
determine the protocols and modes to be included in the testing regime. This will require liaison with the
clinical users to ascertain the primary clinical functions of the system, and the protocols and modes rou-
tinely used for this work – the testing regime should focus on these protocols. It is also important to test
variations due to FoV changes and dose levels.
Abrupt changes in radiographic parameters due to thickness increments can happen and should be prop-
erly captured. Information on the AEC curve (see an example in Figure 1) can be obtained from the manu-
facturer prior to installation to establish which thicknesses should be used when characterising the curve.

Figure 1: Exposure parameters as function of PMMA thickness. The illustration shows how four parameters vary with increasing
PMMA-phantom thickness; green: tube voltage (kV), red: added copper filtration (mm), yellow: tube current (mA) and white:
pulse width (ms).

A.5.4 Dose indicators

The main dose indicators displayed by dynamic x-ray systems are Air Kerma-Area product (PKA), and PKA
rate, and Reference Air Kerma (Ka,r ) and Ka,r rate.
The PKA display is derived either from an ionisation chamber integral to the collimator housing or is calculat-
ed by the x-ray unit using the field size and generator settings and can be displayed in many different units.
PKA is an important patient dose metric and therefore must be calibrated.

version 01.2024 15
In interventional radiology or use of mobile c-arms in theatre, patient views may be taken in an anterior-pos-
terior (AP) projection or in a lateral projection – clearly in one case a patient support will be in the beam and
in the other no patient support will intercept the beam. The PKA meter cannot be set-up to reflect both sit-
uations at the same time. Therefore, it is advised (AAPM TG190) that an ‘in-air’ calibration is performed for
baseline metric determination and Acceptance testing reasons. To supplement this approach, knowledge of
the table support and any mattress/pad support attenuation factors will be required to ascertain patient PKA
for projections that intercept the patient support/mattress. However, the MPE may decide to evaluate such
systems with the table/mattress in the beam if they feel that this is more appropriate. For fixed fluoroscopy
units with an integral undercouch tube the patient support will always intercept the radiation beam and so
evaluation should be made with the patient support in situ. Knowledge of the manufacturer’s calibration
method will aid the interpretation by the MPE of the results obtained.
The AAPM methodology presented here is not the only method to field calibrate a PKA meter/calculation
and attendant display. The IAEA TRS457 Code of Practice also mentions the tandem PKA meter approach
which is described in detail by Toroi et al (2008) and Malusek et al. (2014) This approach utilises a calibrated
PKA meter to directly compare the field PKA meter and reference PKA meter. If the user prefers the ‘tandem’
method, please refer to these references above.
For both approaches one may arrive at an average calibration coefficient across all clinically used beam
areas and qualities or calibration curves depending upon beam area and beam quality. Which approach is
used is dictated by the requirements locally for patient dosimetry precision (if any) and is beyond the scope
of this protocol.
During the testing the displayed P(t)KA must also be compared with that measured.
It should also be noted that some integral chambers measuring PKA are not adjustable, nor is the PKA deter-
mined by calculation within the fluoroscopy unit. For these situations it is unlikely that the PKA relevant to
a given procedure is the displayed PKA and local decisions regarding adjustment must be made. For those
PKA meters that are adjustable, it may be prudent to adjust to have the closest possible agreement (i.e.
calibration coefficient closest to 1.0) with the reference PKA (beam area or tandem method) for the clinical
field size and beam quality most often employed.
For constancy testing, any suitable, time-efficient arrangement will suffice.

A.5.5 Skin Dose Map

Monitoring the skin dose is an essential step for managing the risk of radiation induced tissue reactions. Due
to the variability of the irradiation conditions during interventional procedures, predicting and estimating
the distribution (or map) of skin dose and the maximum value (or peak skin dose) is challenging. Skin dose
mapping (SDM) software products are available for that purpose.
The skin dose is defined as the absorbed energy to the patient skin. It should not be mistaken for the air
kerma at the patient entrance reference point (Ka,r), the position of which is defined in the manufacturer
documentation. Although in some conditions the Ka,r displayed on the angiography unit might be a reason-
able approximation of the skin dose, this is usually not the case (among other reasons, because the Ka,r
does not take into account the x-ray tube position, the patient anatomy, the backscatter generated from
the patient or the table movement). In addition, the skin dose should be reported as a dose to skin tissue
while the Ka,r is calibrated in air. This is of limited effect for the beam qualities encountered in interventional
procedures.
Some vendor-specific SDM software are integrated in the x-ray units informing the clinician as to the pro-
gression of the Peak Skin Dose (PSD) value and location in real time during the procedures. Other SDM
applications generate a skin dose map offline, after completion of the procedure. These are either com-
mercial products included in dose management software or stand-alone software. A review of numerous
products is available in Malchair et al 2020. There is currently no regulatory requirement for x-ray systems
to be equipped with SDM software in Europe as opposed to PKA or Ka,r. However, the most recent version of
IEC 60601-2-43: 2022 introduces the notion of a ‘Dose Map’ integral to the interventional x-ray unit.

16 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Although most SDM software products available in clinics use similar analytical calculation algorithms
(Jones, 2011; Bednarek et al; 2011; Bordier et al; 2015; Gardavaud et al, 2018; Habib Geryes et al, 2018), there
exist significant differences in the implementation that can lead to discrepancies of several tens of percent
in the dose estimates (Dabin, 2021). It is therefore of the utmost importance for the user to have knowledge
of the SDM product calculation algorithm and the parameters customizable to the local use.
The dose map calculations, either online or offline, lend themselves to two types of testing.
The first test is analogous to Acceptance testing: it consists of verifying that, for a given set of input pa-
rameters (e.g., varying angulations), the SDM product gives results consistent with the manufacturer model.
Some manufacturers may provide test procedures that should be followed if available. Otherwise, the user
will have to establish their own testing protocol. The user should therefore have a sound knowledge of their
SDM product in order to design the test methodology.
The second test is analogous to a Commissioning test and evaluates the agreement between the dose val-
ues as reported by the product and the actual PSD to the patient under consideration. The purpose of Com-
missioning SDM technology is to verify how the displayed values of skin dose relate to real-world patients.
Accurate measurements of dose distribution and PSD are challenging and time-consuming. Radiochromic
films can be used for assessing both the accuracy of the PSD and the dose map. However, film dosimetry
can be subject to considerable uncertainty and demands considerable set-up and readout resources. The
calibration for the energy range encountered in interventional procedures is cumbersome, and all elements
from the dosimetry system (not only the films, but also the scanner/densitometer) need to be properly
characterised. Point-like dosimeters might be a more accessible solution to most users, but they cannot be
efficiently used if the position of the PSD is not known in advance (which is often the case on real patients),
nor to verify the accuracy of the dose map.
These tests do not need to be performed as part of regular quality control checks unless stipulated by the
manufacturer.
In the frame of the European-funded project VERIDIC (Validation and Estimation of Radiation Skin Dose
in Interventional Cardiology), up to 10 SDM products were tested on four interventional X-ray systems fol-
lowing a common protocol composed of simple irradiation conditions and one more complex combination
closer to clinical conditions (Dabin et. al., 2021). The PSD estimates generally agreed with the measurements
within ± 40%. Furthermore, the accuracy of the PSD estimate for lateral irradiations could be very poor (up
to 66% underestimation).
Although testing the accuracy of SDM products is challenging and the results should be considered with
caution, this should not prevent MPEs from using SDM information in the daily clinical practice, as SDM
software tools are still expected to be more efficient than only relying on the conventional dose metrics (PKA
and Ka,r) as alert levels triggering patient follow-up for skin reaction.

A.5.6 Detector
This section briefly describes the physical parameters used to characterise an x-ray imaging detector and the
steps required for their measurement in the X-ray room. A standard evaluation includes the response func-
tion, the pre-sampling modulation transfer function (MTF), the noise power spectrum (NPS) and the detective
quantum efficiency (DQE). There is a large body of literature describing the theory (Cunningham, 2000) and
measurement (Dobbins et al. 1995; Samei et al., 2006; Dobbins et al., 2006) of these parameters. This can
be explicitly for the image detector (IEC, 2008) or for the complete x-ray imaging system (Kyprianou et al.,
2005), where aspects such as the focus size influence the system MTF and system DQE. The advantage of
these physical characterization measurements is that performance characteristics of the image detector are
assessed directly and can be tracked over time. This can be contrasted with a technical image quality meas-
urement such as low contrast detectability, where image detector and x-ray tube performance are combined
with many other system aspects including x-ray energy, AEC operating point, recursive temporal filtering and
x-ray anti-scatter grid etc. Those unfamiliar with the theory and measurement should review the literature in
order to understand the strengths and weaknesses of these methods before attempting measurements.

version 01.2024 17
While it is not practical or even possible to apply the International Electrotechnical Commission (IEC)
standard for dynamic imaging systems as a Constancy QC procedure (IEC, 2008), some brief discussion
is instructive. The aim of these protocols is to standardise the evaluation of the imaging properties of
image detectors, under carefully controlled imaging conditions. The x-ray energy and exposure at the
image detector are set precisely in order to achieve this. Furthermore, for dynamic image detectors, the
manufacturer will have full control over the many imaging modes available, which in turn will set important
factors such as the pixel binning and the image detector gain. The fluoroscopy and acquisition programs on
the system will utilise a specific image detector mode, chosen by the manufacturer, that is adapted to that
specific clinical imaging program. The difficulty for MPEs responsible for Commissioning and QC tests lies
in establishing the image detector modes being set for the tests/evaluation and also in setting these same
modes for subsequent QC visits.
Application of these methods hinges upon the availability of images with fixed gain and minimal image
processing (just the image detector gain and offset corrections applied, no clinical image processing set).
If possible, at the Commissioning stage, the MPE should establish, with the aid of the engineer and/or
application specialist, how these images (‘For Processing’ or equivalent) can be acquired routinely as part
of the QC process. This may be via the User QC mode, or an alternative would be to copy a commonly used
acquisition program or fluoroscopy mode and set minimal image processing – this would be done with the
help of an engineer. An advantage of doing this is that the image detector will be assessed in a gain mode
that is relevant to a commonly used imaging mode. A further requirement is that the user has some control
over the x-ray parameters so that the energy can be set for the evaluation and the tube load (mAs) adjusted
so that the x-ray exposure at the image detector can be varied. The extent of the characterization will
depend on access to the different image detector gain modes, the range of FoVs and use of image detector
binning, and the number of focus sizes available.
The image detector response function is a crucial measurement from which detector brightness uniformity,
polynomial noise decomposition and noise power spectrum can be calculated.
As with all QC measurements, it is important that the same imaging conditions are set each time so that any
changes can be isolated i.e., same energy, FoV (image detector binning) and focus. Ideally the same gain
mode would be set at each QC visit, as the MPE would then know what magnitude of x-ray exposure to set
in order to acquire an image set that gives relevant information on the image detector. A high gain mode
with high x-ray exposure would lead to saturation of the image detector, while a low gain and low x-ray
exposure could result in images that contain a lot of electronic noise. Neither of these scenarios would give
much useful information on how the image detector is performing when used clinically. While MPEs have
a reasonable amount of experience applying these methods to mammography and general radiography
image detectors, implementing these methods in an efficient and relevant manner for fluoroscopy and
angiography systems is a fresh challenge.
The IEC protocol (IEC, 2008) implements a lag effect correction factor that is used to correct the NPS and
thus gives a lag-corrected DQE. This is considered outside the scope of this protocol and is therefore not
discussed.
Other than for detector characterisation, these metrics could be used for system characterization using a
different geometry. Imaging the edge at the patient position on the tabletop, instead of at the detector, will
provide a system MTF.
It must be emphasised that although the parameters evaluated in this section have a major influence on the
ability of the operator to perform a given imaging task, this is not an ‘image quality’ section. The MTF and
NPS can be used to characterise the sharpness and the noise present in the images however many more
parameters will determine the task performance, including the x-ray spectrum, the lag (recursive temporal
filtering) and probably the final imaging processing set.
Physical image quality parameters provide insight into the signal to noise ratio transfer of the image de-
tector, which is crucial in determining operator task performance. It is therefore assumed that using the
MTF and NPS to track changes in detector or system performance is of value in QC measurements. Given

18 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

that the MTF and NPS have been measured, it is possible to calculate the DQE but the limitations of the
evaluation must be recognized (i.e. the requirement to evaluate the lag, exact energy and incident air kerma
at the detector surface). To track changes in detector performance it is sufficient just to measure/track the
MTF and NPS (Mackenzie et al., 2010). The exact measurement conditions for the NPS and MTF must be
noted carefully for a given QC visit, especially the applied recursive temporal filtering which can strongly
influence the magnitude of noise in the image.

A.5.7 Image Quality

The images generated by an imaging system are used to perform some specific imaging tasks and must
therefore contain sufficient information for the operator to perform the task, within some level of confidence
(ICRU Rep. No. 54., 1996). The degree to which an imaging system is delivering relevant and sufficient
information to the user is established in a measurement of ‘image quality’. For this protocol, we distinguish
between clinical and technical image quality evaluation: both relate to an assessment of the information
contained within the images, but clearly have a different scope. This section also gives some background
context for the different image quality methods.

Clinical image quality

Quantitative evaluation of task performance
To illustrate the evaluation of clinical image quality, then consider a classification task, where the observer is
shown a set of images, some of which are normal, and some contain pathology. The observer has to assign
each image to one of two classes: a normal case (signal absent) or abnormal case (signal present) (Metz, 1986;
‘ICRU Rep. No. 54.’, 1996). Binary classification of images in this way amounts to performing a detection task.
In doing this, the observer is assumed to generate some internal response that is related to the presence or
absence of the signal of interest; this is termed the ‘decision variable’ and is used to classify each image. The
two samples of normal and abnormal images will produce a range of responses in the observer i.e., generate
a range of decision variables, as seen in Figure 2. In general, the distributions of decision variables will overlap.

Figure 2. Probability density distributions of a radiologist’s confidence in a positive decision for a given diagnostic task, for
a set of images with and without pathology. The different shaded regions show the true positive (TP), true-negative (TN),
false-positive (FP), false-negative (FN) region. The vertical line t shows a setting of the decision criterion; this separates
‘negative’ decisions from ‘positive’ decisions.

version 01.2024 19
When making a decision, the observer adopts some value of the decision variable as a decision threshold;
decision variables above this threshold are cases considered to be abnormal, while those below are con-
sidered normal by the observer. It is clear that unless there is complete separation of the two distributions,
there will be instances where cases are declared abnormal when they are actually normal (the false positive
fraction (FPF)) and cases said to be normal when in fact they are abnormal (the false negative fraction (FNF)).
Correctly identified abnormal cases are given by the true positive fraction (TPF) while those correctly iden-
tified as negative are the true negative fraction (TNF), and are synonymous with sensitivity and specificity,
respectively. Specifying system performance must be done using at least one (sensitivity, specificity) pair, as
just using the TPF does not characterise the system performance with respect to actually negative patients
(Metz, 1986). In fact, during the reading study, the observer is asked to give a rating reflecting their confidence
that a given case is normal or abnormal. This generates sets of (sensitivity, specificity) pairs which can be
plotted to give the receiver operating characteristic (ROC) curve. The different points on the curve are effec-
tively obtained by the observer varying their “decision criterion”. The ROC curve clearly depicts the trade-off
between sensitivity and specificity. Depending on the overlap and spread of the distributions, increasing the
decision threshold to increase sensitivity generally comes at the expense of more false positive decisions.
ROC analysis is often applied to studies involving detection tasks, where a ground truth regarding the
presence or absence of a lesion can be established by another means, for example using histopathology
of lesions, where cases can be collected and used retrospectively in the ROC study. However, applying this
method to angiography or fluoroscopy tasks is particularly challenging. Examinations may, for example,
involve detection of some stenosis or bleeding within a vessel, along with an estimation of the severity
of the case. Active therapeutic intervention will follow, with the placement of a device such as a stent
or balloon. Retrospective use of these image sequences in an ROC study is not possible and hence even
quantifying clinical image quality for dynamic imaging exams using standard methods remains a challenge.
The enormous range of tasks for which dynamic x-ray imaging systems are used further compounds the
problem. In effect, clinical image quality is indirectly measured all the time, during the imaging of patients.
Formal quantification of clinical image quality in studies involving fluoroscopy/angiography will only be
rarely done but this should not prevent the MPE from discussing the clinician’s impression of image quality
when problems arise, reviewing cases together and noting which are considered problematic. The MPE can
then take steps to investigate some specific aspects of technical image quality, generally using test objects
to identify the point(s) within the imaging chain that are limiting clinical quality.
The following sections describe some of the methods available to MPEs for use in Constancy measurement
of technical image quality, along with some historical background information and a word on the strengths
and weaknesses of each method.
Subjective evaluation using visual grading methods
The use of image criteria to evaluate the quality of image sequences in image guided interventions is rare,
compared to their usage in computed tomography, and projection radiography examinations, which has
been formalised in EU publications (Carmichael et al., 1996; Menzel, Schibilla and Teunen, 1999). The lack
of published task-specific image criteria for the large number of different procedures performed using
dynamic imaging may be a reason, further compounded by the complexity of the procedures. There are
some practical advantages of using image criteria evaluation compared to ROC-type evaluation in that the
ground truth need not to be known and therefore basically any stored patient examination can be used in
the evaluation. The task is to assess the visibility of relevant anatomy rather than subtle pathology. However,
the validity of image criteria evaluation, i.e., visual grading methods, relies on a positive correlation between
the visibilities of such criteria and detection of pathology or the ability to safely perform the procedure.
Since the image criteria are often assessed using a graded scale (e.g., 5-point Likert-type) it is important
that the correct statistical methods are used in the analysis. This includes Visual Grading Characteristics
(VGC) (Båth and Månsson, 2007) or Visual Grading Regression (VGR) (Smedby and Fredrikson, 2010)
where the ordinal properties of the dependent variable (graded quality score) are considered rather than
performing parametric tests of means of pseudo-interval Likert-type scores e.g., 1-5.

20 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Attempts to optimise the quality of the images of an image guided intervention by dividing the whole procedure
into stages (Lundh et al., 2021) have been made, where each imaging event is linked to a medical task (e.g., loca-
tion of stent graft or visualise graft and vessels etc.) and an imaging purpose. Assigning specific image criteria
to each stage with its specific imaging purpose may be a way forward to optimise these complex procedures.
Technical image quality, TIQ
Technical image quality tests are important for the evaluation of radiological equipment and use phantoms
to represent the patient.
Four methods have been selected and described in this protocol. Three methods are quantitative methods:
Signal-to-noise ratio rate (SNR2rate) using model observer analysis, threshold-contrast detail detectabil-
ity (TCDD) based on a statistical method and TCDD using model observer analysis. Before implementing
any specific method, the reader is strongly recommended to study the references (see specific protocol
sections). The last method is based on visual scoring, because medical physics services may not have ex-
perience in using quantitative methods. In the long term, a switch to quantitative methods calculated from
images is recommended, as this is more reproducible and more readily compared against a TIQ standard.
In general, the system generates and stores images with a given clinical image processing set applied and
these are usually referred as ‘For Presentation’ images. The characteristics of the image processing will
depend on the acquisition or fluoroscopy program used. When evaluating or discussing clinical image
quality then clearly ‘For Presentation’ images will be used. For the evaluation of technical image quality
with one of the four methods described in this protocol, then physicists will, in general, only have access
to ‘For Presentation’ images from the system. Technical image quality will therefore be routinely assessed
from ‘For Presentation’ images, and physicists must exercise some caution when interpreting these results.
Furthermore, it may also be possible to use ‘For Presentation’ images when calculating Fourier metrics as a
means of tracking detector or system performance over time. The use of image quality metrics calculated
from ‘For Presentation’ fluoroscopy and angiography images for QC purposes is still in its infancy. It will be
some time before sufficient data are available for a broad range of systems and imaging applications to de-
termine whether the results are sensitive and reproducible. When explicitly assessing detector performance
using Fourier methods, access to ‘For Processing’ images is required i.e. images without the clinical image
processing set applied. Access to these image types will generally be via the XR27 user QC mode, or by a
program set up to have fixed gain (no auto-ranging) and minimal clinical processing.
Before applying any of the methods, discuss the procedures performed on the system with the relevant
clinical staff in order to establish the TIQ test settings (e.g., commonly used imaging protocol(s), dose
modes and FoV).
If protocols are amended such that TIQ would be affected, then re-commissioning is required, and new
baseline values established. Situations where the protocols are amended include clinician requests, soft-
ware and/or hardware updates, and new clinical applications.
Detector vs system evaluation: influence of geometry
Broadly speaking, two geometries are used for TIQ testing, depending on the focus of the evaluation. These
are depicted in Figure 3. The detector geometry is easily applied to mobile C-arm type systems used in oper-
ating theatres where access to a table needed to support the scattering material can sometimes be difficult.
The x-ray detector is one of the basic determinants of the quality of the images produced by an imaging
system, setting a fundamental limit on the sharpness and noise of the images that can be produced. For
many years, there was considerable focus on a direct assessment of the x-ray detector in fluoroscopy (Hay
et al., 1985), reflecting the use of x-ray image intensifiers (XRII) coupled to a television camera and QC
testing with a ‘detector’ geometry (Figure 3a). These are vacuum electro-optical devices and can suffer
from a gradual reduction in gain over time, which could be quantified via the conversion factor (Holm
and Moseley, 1964). There could also be a loss of sharpness, due to changes in the focusing voltages and
therefore QC methods tended to focus on detecting changes in the detector (i.e., the XRII-TV system) over
time (Hay et al., 1985). XRII devices have since been largely superseded by digital flat panel x-ray detectors

version 01.2024 21
in fluoroscopy and angiography systems that do not have active electrostatic focusing, although many QC
programs have not adapted to this change. As far as we have been able to, this protocol tries to adapt to
these technological changes as far as practical.
The two principal means of evaluating x-ray detector imaging performance are via Fourier metrics and
using low scatter planar test objects imaged at the detector input plane (Hay et al., 1985; Cowen, Haywood,
et al., 1987) (Figure 3a). In both cases, an appropriate metal filtration is positioned at the x-ray tube port, to
reduce the quantity of scattered radiation produced and giving a defined x-ray beam energy distribution
for the evaluation. Detector input exposure levels can be quantified and the influence of geometric blurring
from the x-ray focus is reduced by positioning the test tools at the detector input plane. As a result, a
clear picture of x-ray detector sharpness, noise and efficiency is obtained, and these can be compared
against reference data and also used to track detector performance over time. However, many important
system parameters which influence clinical image quality and task performance are ignored by this kind of
detector-centric evaluation.

Figure 3. a) geometry for explicit evaluation of the x-ray detector (can also be orientated with x-ray at the bottom/detector
at the top). The anti-scatter grid has to be removed (if possible). b) geometry for measurement of x-ray system performance

If an overall system performance test is required, a ‘system geometry’ has to be used. Many more system
parameters will influence the TIQ score in this geometry, including geometric focus blurring, scattered
radiation and the influence of the anti-scatter grid and the x-ray energy. A detailed evaluation of the x-ray
detector could be made at the Commissioning stage and explicit detector evaluations only made if prob-
lems with the x-ray detector are suspected at some point during the lifetime of the detector. Constancy
TIQ testing of the overall system would be made using test objects in a system geometry. Table 2 lists the
expected PMMA thickness ranges for four common clinical applications and the FoVs used. At Commission-
ing, TIQ can be assessed for a number of patient thicknesses and FoV. Longitudinal testing should track the
commonly used imaging protocol and the associated fluoroscopy mode and acquisition program (series
imaging). This should be done at a relevant phantom thickness for one FoV. It is important to have a relevant
dosimetry measurement to accompany the TIQ measurement. The entrance surface air kerma rate (ESAKR)
should already have been measured at the corresponding phantom thickness(s) in the AEC tests and these
values can be used with the TIQ data, provided the phantoms do not change the beam load/dose.

22 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Table 2. Ranges of patient equivalent thicknesses (PMMA) and field of view (FoV) for different clinical applications. Also
angulated positions are considered.

Clinical application PMMA Thickness (cm) Field of View (cm)

Head 16 cm 16-22 cm
Abdomen 20,25,30 cm 32-42 cm
Cardiac 20,25,30,40 cm 20-25 cm
Paediatric 5-20 cm FoVs vary widely depending upon size and application

Phantoms
An overview of phantoms used in clinical practice is discussed in Appendix 2. Often these phantoms bear
little resemblance to patients and should therefore not be used to optimise procedures since they provide a
too simplistic image representation (e.g., no anatomical varying background). Image processing, which is an
important component of dynamic imaging systems, is likely to have a different response to current TIQ test
objects compared to scenes containing anatomical information.
Current test objects are still typically formed from circular details arranged in a geometric pattern, set within a homo-
geneous background and are usually static when imaged. Extrapolation of the TIQ measurement to clinical image
quality is therefore difficult. A good example of this occurs when using static test objects to assess cardiac imaging
modes, where clinical image processing will be set up for good temporal resolution but at the expense of increased
quantum noise in the images. The use of a static phantom in this situation would give a misleading result. Note that the
implementation of advanced analysis methods would not resolve the problem in this case. Test object composition
and the method used for imaging can also give potentially misleading results. Using copper attenuation at the x-ray
tube to mimic patient attenuation may distort the measured technical image quality performance compared to us-
ing a more realistic attenuator such as PMMA. A test object containing iodine could produce higher SDNR results at
higher tube voltages when imaged with copper, while the use of PMMA attenuation would produce higher SDNR at
lower tube voltages. Furthermore, the use of materials such as aluminium in test objects to optimise protocol selec-
tion must be done with caution. Aluminium will not correctly indicate the optimal setup for objects containing iodine,
steel (stents or guidewires) or any of the higher atomic number materials often used to increase device visibility.
However, these simple test objects form a controlled input that is suitable for constancy testing and can be
used to detect deviations from the imaging performance established at Commissioning.

Signal-to-noise ratio rate – Model Observer

Several papers have described technical image quality evaluation in the spatial domain using statistical
model observers for fluoroscopy or dynamic imaging systems (Tapiovaara, 1993; Tapiovaara and Wagner,
1993; Favazza et al., 2015; Bertolini et al., 2019; Villa et al., 2019). There has been strong growth in the
development and application of these methods in recent years; further information on model observer
methods is available in many papers or reference books (Barrett et al., 1993; Eckstein, Abbey and Bochud,
2000; Abbey and Barrett, 2001; Barrett and Myers, 2003; He and Park, 2013).
A model observer method is a means of applying a mathematical model, usually developed using signal de-
tection theory, to a set of images with the aim of estimating task performance for some theoretical (‘model’)
observer (Barrett et al., 1993). While estimation tasks can be modelled, much of the literature focuses on
detection tasks i.e., classification into signal present or signal absent images. This clearly has parallels with the
measurement of clinical image quality using ROC analysis with human readers. The SNR resulting from these
methods is a measure of the image quality: a well set up system will generate a large separation between the
distribution means and/or reduce the standard deviation on the distributions.
A Model Observer called the DC-suppressing observer (DCS) has been applied to fluoroscopic data (Tapio-
vaara, 1993; Tapiovaara and Wagner, 1993). Its advantages in terms of precision compared to visual threshold
contrast tests or alternative forced choice tests are described in Tapiovaara and Sandborg (2004) in a direct
comparison. This model observer excludes the DC-component (average image brightness) and is constructed
by averaging a large number of images (e.g., 1024) with and without the test object (e.g., a small disc) being

version 01.2024 23
present. The difference of these averaged images Δg is used in computing the decision variable DDCS(g)
where gi,j is the pixel values of each analysed image with the test object and P is the number of pixels in
the imaged area (Tapiovaara, 2003). The model observer template Δg is then cross-correlated with each
image frame separately with and without the test object to form the observer’s decision variable. The SNR
is calculated from the conditional distributions:

where:
- D(g|signal) and D(g|background) are the decision variables
- ‘background’ represents the situation without the test object (see Figure 4)
- ΔD is the standard deviation of the distributions.

Figure 4 An example of the experimental distributions of DDCS for the signal-present (▪) and absent (▫) cases (from Tapiovaara
1993). The Gaussian fits are shown as full curves. The SNRsingle frame in this example is 6.6 and the experimental conditions are 30
cm acrylic phantom, 65 kV, total filtration 2.5 mm AI, no grid, air kerma rate at phantom surface 3.8 mGy min-1.

It is, however, the accumulation rate of SNR 2, i.e., SNR 2rate which is of interest in fluoroscopy. Neighbouring
frames in a sequence are not independent, and hence a lag-factor, F, is calculated from the spatial-temporal
noise power spectrum to account for the number of independent frames per second such that SNR 2rate =
SNR­* F, for details see (Tapiovaara, 2003).
Application of SNR 2rate in Constancy QC is described in (Elgström, Tesselaar and Sandborg, 2021) and in
(Tesselaar and Sandborg, 2016). Longitudinal testing of a fluoroscopy system in a simple QC setup over 4
months showed that uncertainty in SNR 2rate was ±14% (± 2 standard deviations).

Threshold-contrast detail detectability (TCDD) – Statistical method

Recent work by the AIFM (Associazione Italiana di Fisica Medica) describes the use of the statistical method
(SM) for the QC of fluoroscopy and angiography systems (Paruccini et al., 2021). The method was initially
presented by (Chao et al., 2000) for CT imaging, and is closely related to the model developed by (Rose,
1948). The approach is based around the extraction of n ROIs of identical size in pixels (e.g., d x d pixels)
from a homogeneous image region. The mean pixel value of each ROI is calculated (χ) and the standard
deviation of these means (σ χ) is used to determine threshold contrast. The distribution of means is assumed
to follow a normal distribution and the threshold contrast for the statistical method (CTSM) defined for the
95% confidence level:

24 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

This effectively states that there must be some separation between adjacent distributions for an object of
a given ROI size for an object to be considered visible, similar to the threshold visibility condition applied in
the Rose model (Rose, 1948). The size of the ROI has some influence on the correlation of the noise present
within the ROIs. Minimum ROI size is 2x2 pixels, which is of the order of 0.3 mm for a typical detector
pixel spacing of 0.15 mm. The minimum number of ROIs used to estimate σx is taken to be 25 (Paruccini et
al., 2021). Furthermore, the uncertainty of the method depends on the insert size: for smaller inserts, the
coefficient of variation (CV) of C TSM is 2–3%, whilst for the larger inserts (25 subROIs) it may reach 10% or
greater. Repeating this process for a range of sizes generates a contrast-detail curve. An important part of
the method is the use of an aluminium step wedge that allows the influence of x-ray energy on the c-d curve
to be included, by converting pixel value to millimetres of aluminium (see Figure 5).

Figure 5. Statistical Method test object: central homogenous region, aluminium step wedge and lead insert to estimate scatter
and veiling glare effects and image offset.

The method has been adapted to include the influence of temporal integration of the human visual system,
which temporally integrates over ~200 ms, depending on scene luminance. This results in an averaging of
uncorrelated noise in different images and a blurring or temporal smearing of moving signals. To include
this effect on the noise, groups of images that correspond to the approximate eye integration time of
200 ms, e.g., 3 images @ 15 fps, 6 images @ 30 fps are averaged before extracting the ROIs for the C TSM
calculation (Villa et al., 2019).
Implementation requires fewer images compared to model observer methods (Villa et al., 2019). The test
object is inexpensive and robust and can be used standalone, in scatter-free conditions, or with tissue sim-
ulating materials such as PMMA. The method has been applied to both ‘For Processing’ and ‘For Presenta-
tion’ images. It should be noted that this method only uses data extracted from homogenous image regions
containing noise. No detail objects are present and therefore this method cannot assess the influence of
geometric blurring due to the x-ray focus on system spatial resolution, even if the object is positioned at
isocenter. This effect is especially noticeable for small size details (< 1 mm).

Threshold-contrast detail detectability (TCDD) – Model Observer

In order to overcome computational constraints related to spatial model observer methods, many authors
implement a linear observer model called the channelized Hotelling observer (CHO). At the heart of the
CHO method is a template that is typically built and applied as follows:
- Image patches (ROIs) of many images of the expected target (e.g., a disc in a uniform background, a
guidewire or a lesion) are extracted from the set of signal present images along with ROIs containing
signal absent (background) areas.
- These ROIs are multiplied by a set of 2D functions (the ‘channels’) that reduce the dimensions of the data.
The number (n) and shape/type of channels is generally chosen to reflect characteristics of the target.

version 01.2024 25
- Each channel is the size of an ROI and the dot product of the ROI and channel function (e.g., n=1) gives a
scalar value. These scalar values are stacked to form a vector with n elements. The difference between
the signal present and signal absent channelized vectors is defined as the expected signal.
- The covariance matrix is also calculated from the channelized signal present and signal absent vectors.
- The expected signal and covariance matrix are combined to give the CHO template.
- This template is built using a set of signal present and signal absent images in the ‘training’ stage.

Once built, the template (a vector of n values) is applied in the ‘reading’ or ‘testing’ stage, to a fresh set of
signal present and signal absent images. Signal present and signal absent ROIs are extracted, ‘channelized’
i.e., multiplied by the channel set, and then multiplied by the template. This generates a ‘decision variable’
(t), analogous to those assumed for a human reader in an ROC study. Applying the template to many signal
present and signal absent images produces two distributions of decision values (figure 6).

Figure 6. Distributions of decision variables calculated for signal present and signal absent image sets

The signal to noise ratio (SNR) can be calculated using a model observer from the distributions in figure 6
as follows:

where t1 and t0 are the mean value of decision variables for the signal present and absent images, respec-
tively, and σt1t2 is the averaged standard deviation for these distributions.
One of the drawbacks of CHO implementation is the number of images required to obtain a reliable estimate
of the covariance matrix. The rank of the standard covariance matrix is less than or equal to the number of
samples used to estimate the matrix (Gallas and Barrett, 2003), meaning for example that at least 96 images
are needed to form a covariance matrix that can be inverted in a 96 channel CHO template. To obtain an ac-
curate covariance matrix, the number of training images is between 10 and 100 times the number of channels
(Gallas and Barrett, 2003), with a factor of 10 (Favazza et al., 2015) being a typical value. Reducing the number
of channels helps in this respect but comes at the cost of reduced flexibility when modelling different targets.
Spatial domain model observers can be used to produce an objective, quantitative measurement of either
TCDD or low contrast-detail detectability (LCDD) i.e., not at the threshold condition. This is an alternative
to visual scoring of c-d test objects. In the CHO applied by Bertolini (Bertolini et al., 2019), images of the
Leeds TO10 are used, and the output is converted to threshold contrast values. These are then compared
against human reader results generated using a 2-AFC method. Internal noise was added to the CHO results
to obtain a good match between the model and human reader results.

26 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Threshold-contrast detail detectability (TCDD) – Human visual scoring
Until the implementation of analogue to digital converters and digital storage in these systems, it was not
possible to store fluoroscopy images for analysis. TCDD test objects were therefore designed for subjec-
tive (visual) scoring by human readers. As discussed, the arrangement of the discs within the test object
determines how the test object is read out. Figure 2a in Appendix 2 shows an image of the Leeds TO12 test
object and when scoring this test object, the reader has to count to the last disc considered visible. This
type of readout is quick, which is important when testing live fluoroscopy, however there are a number of
drawbacks. TCDD scoring suffers from lack of precision which makes the identification of significant chang-
es in CT difficult in longitudinal testing (Cohen, McDaniel and Wagner, 1984; Marshall et al., 1992; Launders
et al., 1995; Tapiovaara, 1997; Tapiovaara and Sandborg, 2004). There is no way of knowing the percentage
correct scores for the readers in this type of readout, making this method unsuitable for estimating observ-
er efficiency (Burgess, 2011). Perhaps the biggest drawback is the difficulty in defining and communicating
a common detectability criterion among the scorers i.e. what is considered actually visible. As a result, it
is particularly difficult to set standards in terms of CT derived this way and for observers to compare their
scores against this standard.
MAFC is an alternative readout method that requires a different arrangement of discs, where the observer
typically has to state which corner contains the disc, which is then compared against the ground truth disc
position (see Figure 2b in Green and Swets, 1966). The observer has to read several test object images,
enabling CT to be defined at a known percentage correct level. Unfortunately, MAFC scoring is time con-
suming (10 or 15 minutes per image, depending on test object). The determination of a threshold value
(CT) i.e., a signal at the limit of detectability inevitably limits the precision that can be achieved, which is an
important consideration in QC testing (Tapiovaara and Sandborg, 2004). Although suffering from several
weaknesses, visual assessment of test objects using TCDD with ‘count to the last disc’ scoring remains
widely used in fluoroscopy QC programs, largely because of the speed with which the object can be scored.
Limiting spatial resolution (LSR)
The ability of an imaging system to resolve fine details (often termed ‘spatial resolution’) is an important
system attribute for some imaging tasks. The spatial resolution of an imaging system is determined by the
x-ray detector properties and by geometric blurring. In fluoroscopy and angiography applications, geomet-
ric blurring depends on x-ray focus size, the geometric magnification, and any motion between the anatomy
or devices being imaged and the ray focus/detector axis.
The flat panel detectors used in current fluoroscopy and angiography systems are indirect detection devices,
with a needle-like CsI scintillation phosphor coupled to an array of light sensitive detector elements, usually re-
ferred to as pixels. Assuming square pixels, then the pixel pitch (p) can be defined as the distance between the
centre of adjacent pixels in a given detector row. This pitch determines the Nyquist frequency, fN, of the detector:

which is the maximum spatial frequency that can be reproduced by the pixel array without aliasing of the
signal. Ideally, limiting spatial resolution should be assessed with a test object that produces a sinusoidally
varying x-ray signal over a range of spatial frequencies. Practically, a line pair or bar pattern test object is used
to perform this measurement that contains groups of lines each with a specific spatial separation correspond-
ing to a spatial frequency. Groups of rectangular lines are cut into a thin lead sheet and these generate square
waves at a given spatial frequency; a typical line pair test object for fluoroscopy and angiography systems
covers frequencies from 0.5 mm-1 to 5.0 mm-1.
In use, the line pair test object is imaged at a low tube voltage to give a high contrast and at a kerma rate
that limits the amount of x-ray quantum noise in the image (although some noise will be present). The
observer counts to the last linepair group that is clearly imaged and that does not have disturbing patterns
(aliasing) running at some angle across a given linepair group (Albert et al., 2002). This is defined as the

version 01.2024 27
limiting spatial resolution of the system. If the line pair test object is positioned on the x-ray detector
then this result will be close to the Nyquist frequency of the x-ray detector. If the linepair test object is
positioned on the x-ray table then the limiting resolution will be limited by the combined effect of the x-ray
geometric blurring and x-ray detector. Note that if the test object is positioned on the x-ray table, then the
spatial frequencies within each bar group are magnified by the geometric magnification (m) i.e. each spatial
frequency is reduced by a factor 1/m.

A.6. SAFETY

Any new fluoroscopic installation must undergo a comprehensive safety assessment. Including radiation
protection aspects, to ensure the facility is safe for both staff and patients. The assessment includes fa-
cility aspects such as a verification of the shielding of the walls and design layout, as well as procedural
approaches such as radiation risk assessments and optimization of protocols for patient imaging. A com-
prehensive description of these tasks is beyond the scope of a report dealing with fluoroscopic equipment
performance. However, as part of the initial equipment checks there are a number of safety assessments
one may carry out. The performance evaluation of fluoroscopic systems should include also a radiation
safety survey to determine whether the radiological medical equipment, the practical techniques and the
ancillary equipment are complying with the national radiation control regulations, based on the European
Basic Safety Standards (BSS) (Official Journal of the European Union;2014). Also, relevant publications are
the IEC 60601-2-54, last updated in 2022 (IEC 60601-2-54; 2009), and the IEC 60601-2-43, last updated
in 2022 (IEC 60601-2-43; 2010). Practical guidance, especially regarding radiation protection optimization,
has been given by the International Atomic Energy Agency in a 2018 Specific Safety Guide on Radiation
Protection and Safety in Medical Uses of Ionizing Radiation (IAEA; 2018). In many European countries, these
checks will form part of national regulations. Some of these checks are already covered in detail in the
relevant equipment section (e.g., leakage is covered in the ‘X-Ray Tube’ section).
For the staff performing the fluoroscopy procedure in the room, the most important aspect from the ra-
diation safety point of view is the stray radiation (leakage plus scatter) from the patient, the tube, the
image detector and peripheral devices close to the patient. Measurements of stray radiation around the
fluoroscope, using the maximum air kerma rate and the largest field size and with a phantom in the beam
(NCRP, 2010), should be performed before the system is used clinically and after changes in hardware or
software affecting the patient dose (ACR; 2016). Results can be compared with manufactured supplied
isokerma plots.
Details about radiation safety tests are given in Appendix 1.
A succinct list of potential equipment-based radiation protection checks is shown in table 3.

Assessment Potential Methodology

Check for:
Equipment warning lights - Power on lights
- Exposure on lights
Check against national & IEC standards such as:
- Warning alarm after a certain level is reached e.g., 5 min
Equipment alarms of fluoroscopy
- Sound when high dose-rate fluoroscopy is used
- Collision avoidance warnings
Can table be driven to cardiopulmonary resuscitation
UPS functionality
position with mains failure
Check for:
- Correct operation (e.g., ‘dead man’ mode)
Exposure switch/footswitch
-C an exposure be made on leaving the footswitch up-
side down?

28 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 Assess leakage radiation, labelling of focal spot position
and minimum focus-skin distance.
X-Ray Tube
Assess minimum filtration meets national or IEC guide-
lines/regulations
Ensure all blades are just visible and close symmetrically
Collimator
to the centre of the FoV.
Check that the x-ray field and displayed field are aligned
Alignment
within national or IEC tolerances.
Assess Pb equivalence and marking meets national
Pb equivalence of tableside and ceiling shields
standards and specification.

A.7. OTHER GUIDANCE

A number of reports providing guidance on the performance testing and quality control of fluoroscopy
systems have been published over the years. The paper by Boone et al., (1993) reports fluoroscopy expo-
sure rate measurement methods and results from AAPM Task Group No. 11. More general advice on QC of
fluoroscopy equipment by AAPM Task Group No. 12 is given in AAPM Report 74 (Shepard, 2002). More
recent publications are available from the AAPM. Estimation of patient skin dose in fluoroscopy is described
in a joint report by AAPM TG357 and by EFOMP (Andersson et al., 2021). AAPM Task Group Report 272
gives guidance on Acceptance Testing and Evaluation of fluoroscopy systems (Lin et al., 2022).
Fluoroscopy systems are included in the Institute of Physics and Engineering and Medicine (IPEM) Re-
port No 91 that lists standards for diagnostic x-ray system performance testing (IPEM, 2005). IPEM Topic
Group Report 32 Part II provides specific information on the technology and testing of image intensifier TV
systems (Hiles and Starritt, 1996). This has been updated in the IPEM-IOP series report entitled ‘Dynamic
X-ray Imaging Systems Used in Medicine’ (Stevens, 2021). Analysis of fluoroscopy QC results acquired by
MPEs across the UK using methods described in IPEM Report 91 and IPEM Report 32 Part II is described by
(Worrall et al., 2020).
The European Commission has published report No 162 (European Commission, 2012), which provides an
extensive set of non-binding criteria that assesses the acceptability of imaging equipment and advises
on appropriate remedial action, where indicated. These are acceptability criteria and should be seen as
minimum standards rather than remedial levels to be applied to modern, well-adjusted and well-functioning
imaging equipment. The excellent review paper by Jones (Jones et al., 2014) summarises lectures given at
the AAPM summer school in 2012 and covers equipment specification, Acceptance and QC testing plus
information on dose measurement and radiation effects.
The International Electrotechnical Commission (IEC) publishes international standards applicable to fluor-
oscopy systems. These include those that address safety and essential performance (IEC 60601-2-54 &
60601-2-43) and the newly published IEC61223-3-8 (2024) addressing acceptance and constancy testing for
radiography and radioscopy.

version 01.2024 29
SECTION B - PROTOCOL
This section contains the detailed description of the suggested tests. The background to some tests is includ-
ed in Section A.

B.1. MECHANICAL AND GEOMETRICAL PARAMETERS

B.1.1 Determination of source location and minimum source – skin distance

1. I EC 60601-2-43 Medical electrical equipment - Part 2-43: Particular requirements for the basic
safety and essential performance of X-ray equipment for interventional procedures, 2022
2. Accuracy and calibration of integrated radiation output indicators in diagnostic radiology: A re-
Reference
port of the AAPM Imaging Physics Committee Task Group 190 Med. Phys. 42 (12), December 2015
3. IEC 60601-2-54 Medical electrical equipment Part 2: Particular requirements for the basic
safety and essential performance of X-ray equipment for radiography and radioscopy

Tape measure
Instrumentation
Two square test objects, the side of one object is twice the length of the side of the other object

The first action is to ensure that the focal spot position is marked. If it isn't, ask the engineer
to correctly mark the position of the focal spot on the x-ray tube housing.
If the engineer is not present or cannot do this, undertake the following procedure to determine
the focal spot position:
-Use two square test objects, the side of one object is twice the length of the side of the other
object.
Procedure -Place the smallest on the tabletop and attach the largest to the front face of the image detector.
-During fluoroscopy, with the x-ray beam vertical, modify the height of the tabletop or the fo-
cus-to-image detector distance until the two test objects coincide in the image. This will place the
smallest object exactly halfway between the focal spot and the largest object.
-Measure the distance between the two objects. The focal spot position is at the same distance
from the smallest test object.
-Mark permanently on the x-ray tube housing surface for future reference.

Evaluation Compare the measured distance with the reference value

Minimum focal spot - skin distance:

Pass/Fail Criteria -20 cm for surgery application
-30 cm for other application

Frequency Acceptance test

B.1.2 Minimum field size

Tape measure
Instrumentation
Field size test object (e.g. radio-opaque ruler)

Close the collimators as far as possible.

Procedure Place the field size test object in the beam and record the focal spot-to-test object distance.
Acquire an image of the field size test object using a few seconds of fluoroscopy.

Measure both sides of the image obtained.

Evaluation
Rescale the distances to 1 m from the source.

The minimum field size, measured at 1 m from the focus in a plane normal to the reference
Pass/Fail Criteria
axis, shall not exceed 5 x 5 cm2

Frequency Acceptance test

30 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

B.1.3 Beam alignment

1. IEC 60601-2-43 Medical electrical equipment - Part 2-43: Particular requirements for the basic
safety and essential performance of X-ray equipment for interventional procedures, 2022
Reference
2. Radiation Protection n. 162, 2012 “Criteria for acceptability of Medical Radiological Equipment
used in Diagnostic Radiology, Nuclear Medicine and Radiotherapy” 2010
Spirit level
Instrumentation
Beam alignment test tool (plate and PMMA cylinder with two lead marks)
If possible, inspect engineers’ tests results for perpendicularity
If this is not possible, one of the two following methods can be used
Method a):
Rotate the C-arm to a lateral position (90° or 270°)
Select a source detector distance (SID) of 100 cm
Position the test object at the centre of the detector
Acquire an image of the field size test object using few seconds of fluoroscopy

Figure 7. The beam alignment test tool consists of a plate and a cylinder with two radiopaque marks,
placed at its top and at its bottom, respectively.

Procedure Method b):

Place the test object on the couch
Move the couch until the two ball bearings in the cylindrical TO are overlaid on the image (Video
showing positioning of alignment test object)

Figure 8. X-ray misalignment measured in terms of distance between the two radiopaque marks..

Method a):
Evaluate the alignment of the two radio opaque marks
Evaluation Method b):
Allowing for magnification effects calculate the focal offset from centre and convert the offset
into an angle deviation at 100 cm SID
Achievable ≤1.5°
Pass/Fail Criteria
Acceptable ≤3°
Frequency Acceptance test

version 01.2024 31
B.1.4 Correspondence between X-ray field and effective image receptor size

1. IEC 60601-2-54 Medical electrical equipment – Part 2-54: Particular requirements for the basic
safety and essential performance of X-ray equipment for radiography and radioscopy’, 2022
Reference
2. Radiation Protection n. 162, 2012 “Criteria for acceptability of Medical Radiological Equip-
ment used in Diagnostic Radiology, Nuclear Medicine and Radiotherapy” 2010

Tape measure
Instrumentation
External image device (e.g., radiochromic film or CR plate)

Position the external image device in the beam at a known distance to the source (as close as
possible to the source).
Set a FoV and acquire an image using sufficient radiation to produce an adequate image on the
Procedure external image. Make the measurements with the reference axis normal to the image detector
plane within three degrees.
Repeat for all the available FoVs, minimum and maximum SID, vertical (0°) and lateral (90°)
positions.

Correct the image size recorded on the external image device to the SID used for the test.
Verify the correspondence between the nominal FoV and measured x-ray beam size. For rectan-
gular fields, as shown in the figure 9, the measured discrepancies in the image detector plane are
represented by c1 and c2 on one axis and by d1 and d2 on the other.

Evaluation

Figure 9. c1, c2, d1 and d2 assess the correspondence between X-ray and effective image receptor size.

If the image receptor area is circular, the x-ray field shall coincide with the image receptor area
as required in 1) and 2).
1) The x-ray field measured along a diameter in the direction of greatest misalignment with the
image receptor area shall not extend beyond the boundary of the effective image receptor
area by more than 2 cm;
2) At least 80 % of the area of the x-ray field shall overlap the effective image receptor area.
Effective image receptor areas smaller than 10 cm in diameter are exempted.

If the image receptor area is rectangular:

Pass/Fail Criteria
1)Along each of the two major axes of the image receptor area, the total of the discrepancies
between the edges of the x-ray field and the corresponding edges of the image receptor area
shall not exceed 3 % of the indicated focal spot to image detector distance when the image
detector plane is normal to the x-ray beam axis:
|c1| + |c2| ≤ 0.03 x SID
|d1| + |d2| ≤ 0.03 x SID
2)The sum of the discrepancies on both axes shall not exceed 4 % of the indicated source to
image detector distance (SID):
|c1| + |c2| + |d1| + |d2| ≤ 4% SID

Acceptance (SID = min, max and 100cm)

Frequency
Annual test (All FoVs, both angulations and SID = 100cm)

32 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

B.1.5 Verification of displayed distances

1. IPEM Report 91 “Recommended Standards for the Routine Performance Testing of Diag-
Reference
nostic X-ray Imaging Systems” (2005)

Tape measure
Instrumentation
Two square test objects, with one object’s sides twice as long as the other

Vary the SID between the maximum and minimum values (ideally 3 values in total), noting
that the display of SID is equal to the measured value as determined with a tape measure
1) Adjust the table height from the lowest to the highest position and note that the display
Procedure
tracks the difference in height.
2) Record the focus-tabletop distance at zero displayed height and compare with the man-
ufacturer’s reference value.

Compare the measured distances with the displayed data taking into account the nominal
Evaluation distance between the accessible surface of the detector and the detector (as stated by the
manufacturer)

Pass/Fail Criteria The measured distances shall not differ from the nominal distances more than 1.5%

Frequency Acceptance

B.2. X-RAY TUBE AND GENERATOR

B.2.1 Tube voltage accuracy

Aim To quantify accuracy of the set tube voltage on the imaging system

1. IPEM Report 91 “Recommended Standards for the Routine Performance Testing of Diag-
nostic X-ray Imaging Systems” (2005)
2. AAPM report 14 “Performance specifications and acceptance testing for x-ray generators
Reference and automatic exposure control devices” (1985)
3. IEC 60601-2-54 Medical electrical equipment Part 2: Particular requirements for the basic
safety and essential performance of X-ray equipment for radiography and radioscopy
(2022)

Attenuation layers to ensure the tube voltage is set at an appropriate level

Instrumentation
Calibrated multimeter or tube voltage divider

1a. For systems where x-ray factors can be set manually (including user QC mode), set the
tube voltage to a value of approximately 70% of maximum tube voltage achievable e.g.,
80 kV and set a typical loading (mAs) value.
1b. For systems with only AEC mode available, where the tube voltage cannot be set by
the operator, vary the thickness of attenuator positioned between the multimeter and
Procedure
the image detector such that a value of approximately 70% of maximum tube voltage is
achieved e.g. 80 kV.
2. Record the tube voltage, tube load, focus size and spectral pre-filtration
3. Repeat this for at least two other tube voltages, at 90% (~110kV) and at 50% (~60 kV) of
maximum tube voltage value

Evaluation The measured kVp’s shall not differ from the displayed values by more than 8%

Pass/Fail Criteria Calculate the difference between the measured and displayed tube voltage.

Acceptance test
Frequency
Annual test (at least 3 kVp points spread across appropriate clinical range)

version 01.2024 33
B.2.2 Minimum HVL/filtration evaluation

To establish the minimum filtration on the imaging system

Note: The minimum HVL or filtration assessment is carried out for regulatory reasons and the
approach will be determined by the national regulations. For instance, in some jurisdictions
Aim the limit may be given in terms of HVL in which case the measurement may be directly com-
pared with the requirement. However, in some jurisdictions the limit may be given in ‘mm Al
equivalent’ in which case the HVL value measured must be converted into a ‘mm Al equiva-
lent’ value, taking into account the generator waveform, anode angle and kVp.

1. Radiation Protection n. 162, 2012 “Criteria for acceptability of Medical Radiological Equip-
ment used in Diagnostic Radiology, Nuclear Medicine and Radiotherapy” 2010
References 2. IEC 60601-2-54 Medical electrical equipment Part 2: Particular requirements for the basic
safety and essential performance of X-ray equipment for radiography and radioscopy
(2022)

Al sheets of purity ≥ 99%, typically 1 mm thickness

Attenuation layers to ensure the kV is approximately set at 70% of maximum tube voltage
Instrumentation
achievable
Dosimeter or multimeter

1) For systems where x-ray factors can be set manually (including user QC mode), set the
tube voltage to a value of approximately 70% of the maximum tube voltage achievable,
e.g. 80kVp, and set a typical loading (i.e. mAs) value. (Follow manufacturer recommen-
dations to protect the detector from excessive irradiation)
2) For systems with only AEC mode available, and the tube voltage cannot be set by the opera-
tor, vary the thickness of attenuator positioned between the dosimeter and the image detec-
tor such that a value of approximately 70% of maximum tube voltage achievable e.g. 80 kV.
3) Position the dosimeter centrally in the x-ray beam and collimate to the dosimeter and do
not change its position during the test
4) Use an acquisition protocol with minimum filtration (i.e. no added spectral filtration)
5) For systems where x-ray factors can be set manually:
Procedure
a. Record the air kerma for no added Al filter
b. Introduce Al sheets into the x-ray beam, between the x-ray source and dosimeter, re-
cording Al thickness and air kerma each time
6) For systems with only AEC mode:
a. position all the Al sheets in the x-ray beam, between the exit side of the dosimeter and
the image detector
b. record the air kerma for no added Al filter
c. successively remove All sheets from the exit side of the dosimeter to the entrance side i.e.,
between the x-ray source and dosimeter, recording Al thickness and air kerma each time
7) Alternatively, record the half value layer (HVL) and estimated tube filtration given by the
multimeter.

For both system configurations (AEC or manual), determine the HVL using the equation:

where Y0 is the air kerma reading without additional attenuation. Y 1 and Y2 are the air kerma
readings with added aluminium filters thickness of X 1 and X 2, respectively.
If needed, estimate the x-ray filtration from the measured tube voltage and HVL. This can be
Evaluation
done using a spectral model, for example, the Boone model, the SpekCalc or SpekPy models
or IPEM Report 78. For greater accuracy – if the model implementation allows - the target
angle and waveform ripple can be included in the estimate.
Boone Model:
[Link]
SPEKTR 3.0 (Boone Model): [Link]
SpekCalc Model: [Link]
SpekPy Model: [Link]

34 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 Minimum half value layer value shall not be less than the values in the table below.

First half value layer

kV HVL* HVL **
50 1.8 1.5
60 2.2 1.8
70 2.5 2.1
80 2.9 2.3
90 3.2 2.5
Pass/Fail Criteria 100 3.6 2.7
110 3.9 3
120 4.3 3.2
130 4.7 3.5
140 5.0 3.8
150 5.4 4.1

HVL*: Reference values for systems CE marked after 2012 IEC 60601-1-3 2008
HVL**: Reference values for systems CE marked before 2012 IEC 60601-1-3

The minimum filtration must be ≥ 2.5 mm Al or the relevant regulatory limit in place

Frequency Acceptance test)

B.2.3 Half value layer (HVL) evaluation for clinical spectra

Aim To establish the HVL for all the available clinical spectra

1. Radiation Protection n. 162, 2012 “Criteria for acceptability of Medical Radiological Equip-
ment used in Diagnostic Radiology, Nuclear Medicine and Radiotherapy” 2010 IEC 60601-
2-54 (2011)
Reference 2. IEC 60601-2-54 Medical electrical equipment Part 2: Particular requirements for the basic
safety and essential performance of X-ray equipment for radiography and radioscopy
(2022)
3. IEC 61223-3-1 (2001)

Al sheets of purity ≥ 99%, typically 1 mm thickness

Attenuation layers to ensure the kV is approximately set at 70% of maximum tube voltage
Instrumentation
achievable
Dosimeter or multimeter

version 01.2024 35
 1) For systems where x-ray factors can be set manually (including user QC mode), set the
tube voltage to a value of approximately 70% of maximum tube voltage achievable e.g.,
80 kV and set a typical loading (mAs) value. (Follow manufacturer recommendations to
protect the detector from excessive irradiation).
2) For systems with only AEC mode available, and the tube voltage cannot be set by the op-
erator, vary the thickness of attenuator positioned between the dosimeter and the image
detector such that a value of approximately 70% of maximum tube voltage achievable
e.g., 80 kV.
3) Position the dosimeter centrally in the x-ray beam and collimate to the dosimeter.
4) Select the first spectral prefilter to be assessed
5) Set this manually if possible
6) Alternatively, use an acquisition mode that gives this filter setting
Procedure 7) For systems where x-ray factors can be set manually:
a. record the air kerma for no added Al filter
b. successively introduce Al sheets into the x-ray beam, between the x-ray source and
dosimeter, recording Al thickness and air kerma each time
8) For systems with only AEC mode:
a. position all the Al sheets in the x-ray beam, between the exit side of the dosimeter and
the image detector
b. record the air kerma for no added Al filter
c. successively remove Al sheets from the exit side of the dosimeter to the entrance side i.e.,
between the x-ray source and dosimeter, recording Al thickness and air kerma each time
9) Alternatively, record the half value layer (HVL) and estimated tube filtration given by the
multimeter.
10)Repeat for all available pre-filtrations

Calculate the HVL for each spectrum

Evaluation
Estimate spectrum prefiltration

The measured HVL must be consistent with the spectral prefiltration. Possible filtrations
Pass/Fail Criteria
shall be assessed against the manufacturer’s stated values.

Frequency Acceptance

B.2.4 Normalised air kerma

Aim To quantify the Air Kerma normalised to mAs at 1 metre from the x-ray source

1. IPEM Report 32 “Measurement of the performance characteristics of diagnostic x-ray

systems used in medicine” (1996)
Reference
2. AAPM report 14 “Performance specifications and acceptance testing for x-ray generators
and automatic exposure control devices” (1985)

Attenuation layers
Instrumentation Dosimeter or multimeter
Tape measure

1a) 
For systems where x-ray factors can be set manually (including user QC mode), set
the tube voltage to a value of approximately 70% of maximum tube voltage achievable
(about 80 kV) and set a typical loading (mAs) value with no added filtration. (Follow
manufacturer recommendations to protect the detector from excessive irradiation).
1b) For systems with only AEC mode available, where the tube voltage cannot be set by the
operator, vary the thickness of attenuator positioned between the dosimeter and the
Procedure
image detector such that a value of approximately 70% of maximum tube voltage achiev-
able is achieved e.g., 80 kV with no added filtration.
2) Position the dosimeter centrally in the x-ray beam and collimate to the dosimeter.
3) Remove the couch whenever possible, otherwise assess the reduction in the air-kerma.
3) Measure the source-to-dosimeter distance.
4) Record the air kerma, the tube voltage, tube load, focus size and spectral pre-filtration.

Evaluation Calculate the Air Kerma normalised to mAs at a source-to- dosimeter distance of 1 meter.

36 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 Air Kerma normalised to mAs at a source-to-detector distance of 1 meter, 80 kV and approx-
Pass/Fail Criteria
imately 2.5 mm Al filtration shall measure between 25 and 80 μGy/mAs.

Frequency Acceptance test

B.2.5 Leakage radiation

Aim To quantify the Air Kerma normalised to mAs at 1 metre from the x-ray source

1. IEC 60601-1-3: Medical electrical equipment – Part 1-3: General requirements for basic
Reference safety and essential performance – collateral standard: Radiation protection in diagnostic
X-ray equipment. (2021)

Tape measure
Air kerma rate meter with appropriate cross-section area
Attenuating sheets of Pb (lead)
Instrumentation
Note: If an air kerma rate meter of reasonably large size is not available, a dose rate survey
meter (ambient dose rate) sensitive enough and able to respond quickly to changes in the
dose rate could be used.

1) Consult the manufacturer’s test data on leakage radiation and evaluate the content. If this
is not deemed to be sufficient, do the measurement described below.
2) Minimise the x-ray beam and place Pb attenuating layers, e.g., 5 mm, close to the X-ray
tube’s collimator to block the primary beam.
Procedure 3) Select the suitable mode (e.g., high kV) and start x-rays.
4) Move your dosimeter slowly approximately 50 cm from the focal spot and note the dose
rate readings. Measure in all directions and identify the direction with the highest dose
rate. (If your survey dosimeter did not measure air kerma rate (but ambient dose rate),
measure air kerma rate in this direction with a calibrated dosimeter).

Estimate the maximum leakage radiation in terms of air kerma rate at 1 meter from the focal
Evaluation
spot in fluoroscopy mode.

Measured maximum values shall match manufacturer specifications.

When the tube is operating at maximum rating in fluoroscopy mode, the air kerma rate at 1
meter distance from the focal spot must be less than 1 mGy/h.
Pass/Fail Criteria
If the maximum measured air kerma rate at 1 meter is higher than 1mGy/h the clinical activity
shall be suspended. Estimate the position of the possible gap in the shielding and contact
the manufacturer.

Frequency Acceptance test

B.3. AUTOMATIC EXPOSURE CONTROL

B.3.1 AEC function

To test the response of the AEC to varying thicknesses of PMMA in comparison with manu-
Aim
facturer’s data

1. AAPM TG 125 “Functionality and Operation of Fluoroscopic Automatic Brightness Con-

Reference trol/Automatic Dose Rate Control Logic in Modern Cardiovascular and Interventional An-
giography Systems“, (2012)

Tape measure
Instrumentation 35 cm to 40 cm slabs of PMMA 30 cm x 30 cm with thicknesses to allow increment of 1 cm
between 5 cm and 40 cm

version 01.2024 37
 Select a clinically relevant imaging mode for which the manufacturer has supplied data.
Verify if the manufacturer dictates a geometrical set-up for the test and if so follow it.
Otherwise follow the described procedure:

For units with an integrated patient support

1) Position the tube underneath the table.
2) Remove the mattress.
3) Set the table height in order to have the entrance surface of the PMMA phantom at the
reference point [if ESAK/ESAKR is measured simultaneously, support blocks should be
used to support the PMMA in order to be able to introduce the dosimeter below the
stack].
4) Lower the image detector as much as possible.
5) Select a clinically relevant protocol, FoV and mode that you have manufacturer’s data
for. For all thicknesses of attenuator, the relevant exposure parameters that change as a
function of attenuator thickness must be recorded - depending upon the manufacturer
and model these may include the following: kV, mA(s), pulse width, spectral shaping filter,
focal spot and entrance exposure to the detector. If the detector entrance air kerma is
to be measured, ensure the detector is positioned on the detector as per manufacturer’s
instructions or outside the measuring field for the mode in question]. Do for both fluoros-
copy and acquisition.
6) Starting with 15 cm PMMA, add 1 cm each time, repeating the procedure, and once the
exit surface of the PMMA block is within 5 cm from the detector housing entrance, start
Procedure raising the detector to keep 5 cm between the exit surface of the stack and entrance of
the detector housing. For all thicknesses keep the table height fixed.
7) Keep adding PMMA until you reach the tube loading limits for your system. You may wish
to exceed this thickness to ascertain what happens for extremely thick patients.
8) Remember to ensure that during fluoroscopy the factors have stabilised prior to record-
ing, and also undertake a short amount of fluoroscopy before each acquisition to ensure
factors start at the correct value for the thickness in the beam.
For mobile units
1) To support the PMMA either use a patient support similar to that used clinically or posi-
tion the mobile in a lateral angulation and support the PMMA with a trolley.
2) If you are working laterally, take care to position yourself on the detector side of the
set-up, to limit exposure to scattered radiation.

Additional Notes
- For units that undertake paediatric procedures, start at 5 cm thickness.
- For mini C-arms start at 1cm and stop at 10 cm thickness. Repeat the above procedure for
all clinically relevant modes (use the appropriate FoV for the clinical mode).
- If you attempt to measure phantom exit air kerma or detector air kerma, be aware that this
is highly scatter and geometry dependent so repeated measurements may vary greatly
unless extreme care is made to ensure your set-up is exactly the same if retested
- When testing against a manufacturer's curve, you should sample at smaller PMMA steps
(e.g. 1 cm) across discontinuities in defined parameters. Where parameters are slowly
changing, a 5 cm increment in PMMA thickness may be more efficient.

For Acceptance testing, the functioning of the AEC system should be tested for all the clin-
Evaluation
ically relevant imaging modes (e.g., fluoroscopic, acquisition mode) as advised by the MPE.

Where manufacturer data exists, compare the results with those of the manufacturer. Raise
Pass/Fail Criteria
with the manufacturer any deviations that, in the judgement of the MPE, are significant.

Frequency Acceptance test

38 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

B.3.2 Comprehensive evaluation of AEC

To characterise the performance of the AEC for Commissioning and optimisation, and to
Aim
ensure continued acceptable performance.

1. AAPM TG 125 “Functionality and Operation of Fluoroscopic Automatic Brightness Con-

Reference trol/Automatic Dose Rate Control Logic in Modern Cardiovascular and Interventional An-
giography Systems”, (2012)

Dosimeter (two if also trying to measure detector air kerma)

Instrumentation Tape measure
35 cm to 40 cm slabs of PMMA 30 x30 cm

For units with an integrated patient support

1) Set the x-ray unit, remove the mattress, position the dosimeter(s) and phantom as de-
picted in the figure below (Courtesy of AAPM). If you are trying to also simultaneously
measure the phantom exit air kerma or detector entrance air kerma, position an additional
dosimeter in a way that is easily repeatable from year to year.
2) Set the table height in order to have the entrance surface of the PMMA phantom at the
patient entrance reference point (if ESAK/ESAKR is measured simultaneously, support
blocks should be used to support the PMMA in order to be able to introduce the dosimeter
below the stack).
3) Lower the image detector as much as possible.
4) Select a clinically relevant protocol, FoV and mode as advised by the MPE. Acquire an
image and record the exposure parameters: kV, mA(s), pulse width, spectral shaping filter
and entrance exposure to the detector has to be recorded as a function of the attenuator
thickness.
5) Starting with 15 cm PMMA, add 1 cm each time (see note in Frequency below to see how
it changes in annual checks), repeating the procedure, and once the exit surface of the
PMMA block is within 5 cm from the detector housing entrance, start raising the detector
to keep 5 cm between the exit surface of the stack and entrance of the detector housing.
6) Keep adding PMMA until you reach the tube loading limits for your system.

For mobile units

1) To support the PMMA either use a patient support similar to that used clinically or position
the mobile in a lateral angulation and support the PMMA with a trolley.
2) If working laterally, take care to work on the detector side of the set-up.
Procedure 3) The fluoroscopic imaging parameters kVp, mA, pulse width, spectral shaping filter as
well as entrance surface kerma/entrance surface kerma rate at the phantom (and the exit
phantom air kerma/detector air kerma if measured) shall be recorded and measured as a
function of the attenuator thickness

Figure 10. Typical geometries used for system verifications. The vertical one is suitable in case of
units with an integrated patient support (up). The lateral geometry is used for mobile units (down).

Note
- For units that undertake paediatric procedures, use 5 cm thickness.
- For mini C-arms start at 1 cm and stop at 10 cm thickness.
- Repeat the above procedure for all clinically relevant modes (including varying the FoV)
for both fluoroscopy and acquisition.
- When solid state dosimeters are used apply the factor 1.35 to take into account the
backscatter.

version 01.2024 39
 Upon completion of the data acquisition, the fluoroscopic imaging parameters and en-
Evaluation trance (and exit) air kermas should be plotted against the PMMA thickness. The graphs thus
obtained reveal how the imaging system responds to changes in PMMA thickness.

The results obtained are to be used as a baseline following Commissioning and to aid
Pass/Fail Criteria optimisation decisions.
For constancy tests, the deviations from the baseline shall be within 20%.

Commissioning
Annual reduced set of tests (dependent upon MPE judgement and clinical usage)
Frequency - Fix the PMMA thickness and vary the mode and FoV
- Fix the mode and FoV and vary the PMMA thickness [e.g., 10 cm, 20 cm, 30 cm]
- Ensure high usage clinical set-ups are covered.

B.4. DOSE INDICATORS

B.4.1 Air-kerma – Area product and rate display accuracy

To determine the accuracy of the displayed air kerma-area product, PKA and displayed air
Aim
kerma-product rate, P(t)k,a

1. AAPM TG 190 “Accuracy and calibration of integrated radiation output indicators in diag-
nostic radiology” (2015)
2. IEC 60601-2-54 “Particular requirements for the basic safety and essential performance
of X-ray equipment for radiography and radioscopy” (2022)
3. IEC 60580 – Dose Area Product Meters (2019)
4. IAEA TRS457 – Dosimetry in Diagnostic Radiology: An International Code of Practice
Reference (2007)
5. Toroi P et al, A tandem calibration method for kerma-air product meters, Physics in Medicine
& Biology 53 (2008) 4941-58
6. Malusek A et al, In-situ calibration of clinical built-in KAP meters with traceability to a
primary standard using a reference KAP meter, Physics in Medicine & Biology 59 (2014)
7195-210
7. Accompanying documentation and Instructions for Use for the system under test

Attenuating layer
Dosimeter
Instrumentation
Tape measure
Field-size measurement plate (FSMP)

40 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 For interventional and mobile systems in air
1) Set maximum SID
2) Adjust the table height so that it is approximately 50% distance between tube focus and
detector
3) Mount the FSMP in the beam
4) Image the FSMP and adjust the collimators to achieve an imaged field size of approximately
10 x 10 cm
5) Remove the FSMP and place the dosimeter on the central axis of the x-ray beam, without
changing the collimation. The dosimeter can be placed at another point, different from
the FSMP, and then the measured air kerma corrected to the FSMP distance through the
inverse square law
6) Place the attenuators as close to the image detector as possible, ensuring that they fully
intercept the beam, to ensure the operating tube voltage is between 80-90 kVp
7) Noting the PKA display pre- and post-exposure, irradiate the dosimeter in both fluoroscopy
and acquisition mode. Aim for an integrated air kerma in the range 10-100 mGy. Where
relevant, note the air kerma rate during exposure and the displayed P(t)KA .
8) Multiply the integrated air kerma by the area to calculate the measured PKA and P(t)KA and
compare with the displayed PKA and P(t)KA .
9) For Acceptance testing, repeat steps 3-8 for the following variable conditions:
a. Repeat the initial set-up 3 times and determine the coefficient of variation
b. Adjust the attenuator thickness to drive the x-ray system to 50-60 kVp and 100-110 kVp
and repeat the measurements to verify the energy dependence
c. For 80-90 kVp set a field size of approximately 5 x 5 cm and repeat the measurement to
Procedure verify the field size dependence
Note: this test may also be conducted in a lateral beam projection.
Systems with non-removable integral patient support
1) Set maximum SID
2) Remove the mattress
3) Position the FSMP on the patient support
4) Image the FSMP and adjust the collimators to achieve an imaged field size of approximately
10 x 10 cm
5) Remove the FSMP and place the dosimeter at the same position on the central axis of the
x-ray beam.
6) Place the attenuator as close to the image detector as possible, ensuring that they fully
intercept the beam, to ensure the operating tube voltage is between 80-90 kVp
7) Noting the PKA display pre- and post-exposure, irradiate the dosimeter in both fluoroscopy
and acquisition and/or radiographic mode. Aim to achieve an integrated air kerma of ap-
proximately 10 mGy. Where relevant, note also the displayed PKA rate and air kerma rate
given by the dosimeter.
8) Multiply the integrated dose by the area to calculate the measured PKA and compare with
the displayed PKA . Repeat for the P(t)KA .
9) For Acceptance testing, repeat steps 3-8 for the following variable conditions:
a. Repeat the initial set-up 3 times and determine the coefficient of variation
b. Adjust the attenuator thickness to drive the x-ray system to 50-60 kVp and 100-110 kVp
and repeat the measurements to verify the energy dependence
c. For 80-90 kVp set a field size of approximately 5 x 5 cm and repeat the measurement to
verify the field size dependence

Use the ratio of the measured to displayed values to ascertain a calibration factor.
Following successful Acceptance, the MPE should assess whether or not a calibration factor
should be applied to patient PKA values, based upon the calibration factor measured.
Evaluation For constancy checking, if the calibration value differs from the baseline test an MPE decision
should be made regarding further adjustment of patient dose values using PKA .

Note: For systems with non-removable integral patient support, attenuation of the couch
has to be considered.

At Acceptance the PKA value measured shall be within 35% of that displayed across all parameter
Pass/Fail Criteria variations.
For consecutive tests, the calibration factor should not change by more than 5%.

Acceptance testing
Frequency
Annual test (single point)

version 01.2024 41
B.4.2 Manufacturer specified Air-kerma rates

To determine the manufacturer specified air kerma rate and compare to the manufacturer’s
Aim specification.
To check display values

1. IEC 60601-2-54 “Particular requirements for the basic safety and essential performance
Reference of X-ray equipment for radiography and radioscopy” (2022)
2. Accompanying documentation and Instructions for Use for the system under test

30 cm x 30cm slabs of PMMA up to at least 20 cm (check manufacturer’s specification)

Instrumentation
Dosimeter

1) Set up the attenuating layer, geometry and dosimeter position as instructed in the manu-
facturer’s documents
2) Utilise an attenuating layer of 20cm PMMA or as instructed by the manufacturer if differ-
ent from 20cm PMMA
3) Initiate fluoroscopy at the specified equipment settings and geometry
4) Measure the air kerma-rate and compare with the manufacturer’s specification
5) If the manufacturer specifies target Air Kerma per frame for acquisition, at the manu-
Procedure facturer recommended equipment settings, measure the cumulative air kerma and the
number of frames or measure the air kerma-rate and the frame-rate, then divide one by
the other to get air kerma / frame and compare results with manufacturer’s specification
6) Repeat the above procedure for all clinically relevant modes and FoV’s where manufac-
turer’s reference air kermas are provided.
7) If your dosimeter allows for simultaneous air kerma and air kerma rate recording, use the
integrated air kerma from the above tests, adjusted to the Patient Entrance Reference
Point, to compare with the displayed reference air kerma metric.

Compare against manufacturer’s specification

Evaluation
Compare to displayed values

Acceptance - within +/-50% of manufacturer’s specification

Constancy – remedial +/- 20% from baseline

Ka,ref value measured shall be within 35% of that displayed across all parameter variations.
Pass/Fail Criteria
Note: In practice, many units will perform with a constancy such that a 20% deviation from
baseline may be viewed as excessive, in which case a local decision could be made to reduce the
remedial level. It has been set at 20% as the minimum dose change likely to lead to a clinically
detected image quality change.

Acceptance test
Frequency
Annual test for displayed values only

B.4.3 Measurement of table and mattress attenuation

To determine, at acceptance, the attenuation of the manufacturer supplied tabletop and

mattress and to compare with manufacturer supplied values. (This only applies to systems
Aim
with integrated tables, not mobiles, although the MPE may wish to determine the attenua-
tion of non-manufacturer supplied tables etc using a similar method)

Manufacturer specifications
Reference 1. IEC 60601-2-54 “Particular requirements for the basic safety and essential performance
of X-ray equipment for radiography and radioscopy” (2022)

Dosimeter
Instrumentation Tape measure
Attenuators (if using the x-ray equipment in a clinical mode)

42 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 Manual/Service Mode
1) Set kV to 100 kV and ensure a beam HVL of approximately 3.6 mm Al
2) Set for a single exposure of approximately 20 mAs
3) Position the dosimeter such that the tabletop may be moved either between the dosime-
ter and the beam or such that it does not intercept the beam
4) Measure the air kerma with no attenuators in the beam
5) Measure the air kerma with the tabletop in the beam
6) Measure the air kerma with either both the tabletop and mattress in the beam or the mattress
in the beam on its own
7) Calculate the attenuation for both the tabletop and mattress separately and, using a suitable
spectrum generator, calculate the Al equivalence

Clinical Mode
Procedure
1) Select a low filtration fluoroscopy mode
2) Add attenuators to achieve the desired tube voltage [between about 80-100 kVp]
3) Position the tabletop in the primary beam (without the mattress) and place the dosimeter
on the beam entry side of the table in the centre of the primary beam – measure the air
kerma rate from fluoroscopy and the focus-dosimeter distance.
4) Keep the tabletop in the beam and move the dosimeter to the beam exit side of the table
– measure the air kerma rate and the focus-dosimeter distance.
5) Repeat steps 3 and 4 with the mattress on the tabletop
6) C alculate the attenuation for both tabletop and mattress separately and, using a suitable
spectrum generator, calculate the Al equivalence.

Note: for the clinical mode remember to use the inverse square law to adjust the air kerma
rates to a single distance before calculating the attenuation.

Compare the Al equivalence derived for both the tabletop and mattress to the manufacturers
supplied figures.
Evaluation
Ensure that the manufacturers supplied figures are less than or equal to the values in IEC
60601-2-54 Table 203.104.

At acceptance the derived Al equivalence should be within 20% of the manufacturer supplied
Pass/Fail Criteria
figures and less than or equal to the values in IEC 60601-2-54 Table 203.104

Frequency Acceptance test

B.4.4 Limiting Air Kerma Rate

To determine the manufacturer specified limiting air kerma rate and compare to the manu-
Aim
facturer’s specification.

1. National regulations (where such exist)

2. IEC 60601-2-54 “Particular requirements for the basic safety and essential performance
Reference
of X-ray equipment for radiography and radioscopy” (2022)
3. Local Protocol Set-up

30 cm x 30 cm slabs of PMMA up to at least 20 cm (check manufacturer’s specification)

Instrumentation
Dosimeter

1) Set up the test as per B.4.2 (acceptance) or B.3.2 (constancy)

2) Add a Pb apron or similar attenuating layer to the existing PMMA
3) Initiate fluoroscopy at the normal and High Level Control (HLC) fluoroscopy modes, ensuring
Procedure
the radiographic factors have reached maximum value.
4) Measure the dose-rate and adjust the values using the inverse square law to 30cm in front
of the accessible entrance of the detector.

Compare against manufacturer’s set-up and IEC limits.

Evaluation
Note: some equipment may be locally set to have a lower limiting air kerma than the IEC
limits – if so, compare your results to these local values.

version 01.2024 43
 Acceptance: ≤ manufacturer set-up or 88 mGy/min for normal fluoroscopy or 176 mGy/min HLC
Pass/Fail Criteria fluoroscopy mode.
Constancy: as above

Acceptance test
Frequency
Annual test

B.5. SKIN DOSE MAPS

B.5.1 Skin dose map – Acceptance

To verify that the SDM software gives results that are consistent with the model employed
Aim
by the software

Reference Manufacturer specifications

Manufacturer model
Instrumentation
Phantom to simulate a patient

2) Place the phantom on the table.

3) Irradiate the phantom using the following conditions and for each step record the level of
exposure (Ka,r) that is displayed on the system, as well as all the other parameters used
by the SDM software:
Procedure a. A few simple angulations.
b. Angulations that generate beam overlap.
c. Vary table height and lateral/longitudinal positions.
d. Vary beam quality (kV/filtration)
e. Vary FoV.

For each step, calculate what the system would give using the method implemented by the
Evaluation
SDM software and compare with the SDM software display.

Pass/Fail Criteria NA

Frequency Acceptance test

B.5.2 Skin dose map – Commissioning

To evaluate the agreement between the dose values reported by the SDM software and the
Aim
actual PSD

Reference Not Available

Real patients or phantom to simulate a patient

Instrumentation
Dosimeter (e.g., radiochromic films, point detector, grid of point detectors)

Verification of patient dose

1) Place the dosimeter in the area of the patient skin that will be exposed during the procedure

Verification of phantom dose

1) Place the phantom on the table.
2) Place a suitable dosimeter on the phantom’s surface that will be most irradiated during the
Procedure evaluation.
3) Irradiate the phantom using the following conditions:
a. Angulations that generate beam overlap.
b. Vary table height and lateral/longitudinal positions.
c. Vary beam quality (kV/filtration)
d. Vary FoV.

Evaluation Compare measured results with calculated results by the SDM software.

Pass/Fail Criteria NA

Frequency Acceptance test

44 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

B.6. DETECTOR

B.6.1 Response Function

To assess the response of the X-ray image detector in terms of output pixel value as
Aim
incident air kerma/image at the detector entrance is varied.
1. IEC 62220-1-1. Medical electrical equipment - Characteristics of digital X-ray imag-
ing devices - Part 1-1: Determination of the detective quantum efficiency - Detectors
used in radiographic imaging; 2015
2. IEC 62220-1-3. Medical electrical equipment - Characteristics of digital X-ray imag-
Reference ing devices - Part 1-3: Determination of the detective quantum efficiency - Detectors
used in dynamic imaging. Geneva, Switzerland; 2008.
3. Mackenzie A, Doshi S, Doyle P, Hill A, Honey I, Marshall NW, et al. IPEM Report 32
(Part VII) Measurement of the performance characteristics of diagnostic x-ray sys-
tems: digital imaging systems. York, UK; 2010.
Dosimeter
Instrumentation Validated software for calculation of parameters
Filters (1 mm or 2mm Cu or 21 mm Al)
1) Set a suitable mode for physical image quality measurements (e.g., User Quality
Control Mode, or a clinical mode with no clinical image processing applied). The
mode must give the user control over the tube load (mAs) and the tube voltage;
the mode must acquire and store images without clinical image processing (’For
Processing’ or equivalent) and without compression or resizing the images.
2) Note the imaging mode (Fluoroscopy, Cardiac or Series Mode) and detector gain
mode if known.
3) Set the geometry for the physical image quality measurements. Figure 11 shows a
lateral arrangement suitable for measurement of the detector response function and
NNPS
4) Set the SID to represent clinical practice - use the same SID as was used for image
receptor incident air-kerma (IR-IAK) rate measurements. Be consistent with the
previously performed detector response test

Procedure

Figure 11. Typical setup geometry used for detector verifications.

5) Open the collimators fully and ensure the beam shaping filters are not visible
in the image.
6) Position the dosimeter on axis, record the source-to-dosimeter distance and
calculate the inverse square correction to the detector input plane

version 01.2024 45
 7) Remove the anti-scatter grid if possible
8) Set the beam quality for the evaluation:

Radiation Additional Approx mean HVL Photons (mm-2 µGy-1)

kV
Quality filtration energy (keV) (mm Al) i.e., SNRin2
RQA 3 50 10.0 mm Al 40.3 4.0 21759
RQA 5 70 21.0 mm Al 53.3 7.1 30174
RQA 7 90 30.0 mm Al 64.0 9.1 32362
RQA 9 120 40.0 mm Al 77.0 11.5 31077
IPEM 70 70 1.0 mm Cu 56.6 ~7.9 ~32300
IPEM 70 70 2.0 mm Cu 60.1 ~9.0 ~33820

9) Establish the tube current and pulse length settings required for the “normal” level
for the Imaging Mode and also to cover the image detector incident air kerma/im-
age range:

Fluoroscopy Cardiac imaging Series Imaging/DSA

(Imaging Mode 1) (Imaging Mode 2) (Imaging Mode 3)
Procedure
“normal” level 20 nGy/im 200 nGy/im 2000 nGy/im
IR-IAK/image
0 (dark); 0 (dark); 0 (dark);
range for detector
5 to 80 nGy/im 50 to 800 nGy/im 500 to 8000 nGy/im
response
IR-IAK /image 6 nGy, 20 nGy 60 nGy, 200 nGy 600 nGy, 2000 nGy
for NPS and lag and 64 nGy and 640 nGy and 6400 nGy
IR-IAK /image
64 nGy 640 nGy 6400 nGy
for MTF
Tot No. of consecutive
≥128 ≥128 approximately 12
images

10) Remove the dosimeter.

11) At each IR-IAK /image level (tube current and pulse length setting),
acquire the required number of images.
12) Store the images after each fluoroscopy sequence.
13) Repeat i.e., change the mAs and acquire series until the IR-IAK /image range
is covered
14) Repeat for the other imaging modes if evaluating theme.g.,
Cardiac Mode, Series Mode)
15) Transfer the series for analysis (e.g., USB drive or network).
1) For each imaging mode evaluated, there will be a series of images for the response
function, acquired at the IR-IAK /image levels relevant to that mode.
2) C alculate mean and variance of pixel values in a 5x5 mm ROI at the image centre for
the whole sequence to see where the pixel values and variance have stabilised/pla-
teaued. Select the images(s) for analysis from a stable part of the sequence
(as a guide, the first 10 images should be skipped).
3) Measure the mean pixel value (MPV) in a region of interest (ROI) of 5 x 5 mm
Evaluation:
at the centre of the irradiated field.
4) Do this for all the series acquired i.e., all the IR-IAK /image levels
5) Plot MPV versus IR-IAK/image and fit the model function (typically linear, logarithmic
or power). For example, versus air kerma (IR-IAK /image): MPV = A + B × K,
or MPV = A + B × exp(K) or MPV = A + BC
6) Record the fit coefficients (A, B, C etc) and the correlation coefficient for the curve
fit (R 2). Track these over the lifetime of the image detector

46 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 There should be a simple, monotonic relationship between IR-IAK/image and output
Pass/Fail Criteria pixel value.
Expect R² > 0.99
Commissioning: if possible, assess response function for one example each
of Fluoroscopy, Cardiac and Series imaging.
Frequency
Annual: evaluate response function for one fluoroscopy mode (either in User
QC Mode, or for a commonly used fluoroscopy mode).

B.6.2 Detector presampling Modulation Transfer Function (MTF)

Aim To assess sharpness of images produced by the image detector using

the pre-sampling MTF
Reference 1. Cunningham IA. Applied Linear-Systems Theory. In: Jacob Beutel, editor. Medical Imaging
Volume 1 Physics and Psychophysics. Bellingham: SPIE - The International Society for
Optical Engineering; 2000. p. 79–156
2. Samei E, Ranger NT, Dobbins JT, Chen Y. Intercomparison of methods for image quality
characterization. I. Modulation transfer function. Med Phys. 2006;33(5):1454–65
Instrumentation Dosimeter, validated software for calculation of parameters, filters (1 mm or 2mm Cu or 21
mm Al), radio-opaque edge of dimension ≥ 5 x 5 cm, with straight, sharp edges.
Procedure 1) Set a suitable mode for physical image quality measurements (e.g., User Quality Control
Mode, or a clinical mode with no clinical image processing applied).
2) Note the imaging mode (Fluoroscopy, Cardiac or Series Mode) and detector mode
if known.
3) Set the geometry for the MTF measurements
4) Set SID to represent clinical practice (be consistent with previous year).

Figure 12. Typical setup geometry used for modulation transfer function assessment.

5) Remove the anti-scatter grid if possible.

6) Set the beam quality for the evaluation (see B.6.1)
7) Set the tube current and pulse width settings required for MTF evaluation IR-IAK /image
level (see B.6.1).
8) Position the edge test device to measure the horizontal or vertical MTF, as desired.
Rotate the edge to give an angle of 1.5° to 3.0° between the edge and the pixel matrix
columns or rows.
9) If possible, set the smallest focus.
10) Set largest FoV.
11) Acquire the sequence of images.
12) Store the images after each fluoroscopy sequence.
13) Reposition the edge to measure the MTF in the orthogonal direction.
14) Acquire the sequence of images.
15) Repeat for all other FoVs.

version 01.2024 47
Evaluation 1) Linearize the images containing the edge image using the response function (section B.6.1).
2) Calculate the horizontal and vertical MTF curves. It may be necessary to average a number
of MTF curves calculated from different images in the sequence if image noise is high
3) Record spatial frequencies at which the MTF has fallen to 0.5 (MTF0.5) and 0.1 (MTF0.1).
4) Repeat calculations for all FoV.
Pass/Fail Criteria The measured MTF for the image detector should correspond to reference MTF data provided
by the manufacturer, for a given FOV and Imaging Mode. MTF0.5 and MTF0.1 should be within
± 0.2 mm-1 of the baseline value.
Frequency Acceptance test(all FoVs).

B.6.3 Image detector Brightness Non Uniformity (BNU)

To assess the uniformity of brightness across the image detector and check for
Aim
presence of artefacts
1. Mackenzie A, Doshi S, Doyle P, Hill A, Honey I, Marshall NW, et al. IPEM Report 32
(Part VII) Measurement of the performance characteristics of diagnostic x-ray systems:
Reference digital imaging systems. York, UK; 2010
2. Marshall NW, Mackenzie A, Honey ID. Quality control measurements for digital x-ray
detectors. Phys Med Biol. 2011; 56(4).
Instrumentation Dosimeter, validated software for calculation of parameters, filters (1 mm or 2mm Cu or 21 mm Al)
1) Acquire a sequence using the setup for the response function but using the largest
FoV. If doing this, use a clinically relevant IR-IAK/image. Alternatively, use a sequence
obtained for the response function measurement that has a clinically relevant IR-IAK/
Procedure
image for the BNU calculation.
2) This IR-IAK/image level should be used for the BNU evaluation throughout the lifetime
of the detector unless there is a large change in the set (target) IR-IAK /image.
1) Linearize the image to air kerma using the response function.
2) Extract a centrally positioned ROI from the image. The ROI should cover most
of the FoV, leaving a gap of ~1 cm at the image edge.
3) From this large ROI, extract 20 x 20 mm half-overlapping small ROIs.
4) Within each small ROI, calculate the mean pixel value (SI ) of the pixel values.
5) Record the value of (SI ) at the ROI position and repeat for all ROIs extracted
from the large ROI.
Evaluation
6) Calculate the maximum (SI max), minimum (SI min) and average (SI avg) value of the
small ROIs and then calculate the BNU:

7) Examine the image for artefacts, especially for signs of flat-fielding artefacts.
Signal brightness should be uniform across the image; BNU should be <10%.
Pass/Fail Criteria
No artefacts visible.
Commissioning: all FoVs for systems with FoV-specific flat field correction,
Frequency else only largest FoV
Annual test: Largest FoV

48 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

B.6.4 Variance and SNR Image

Aim To assess the global uniformity of the image noise and check for presence of artefacts
1. Mackenzie A, Doshi S, Doyle P, Hill A, Honey I, Marshall NW, et al. IPEM Report 32
(Part VII) Measurement of the performance characteristics of diagnostic x-ray systems:
digital imaging systems. York, UK; 2010
2. Marshall NW, Mackenzie A, Honey ID. Quality control measurements for digital x-ray
Reference
detectors. Phys Med Biol. 2011;56(4).
3. Monnin P, Bosmans H, Verdun FR, Marshall NW. Comparison of the polynomial model
against explicit measurements of noise components for different mammography
systems. Phys Med Biol. 2014;59(19)
Instrumentation Dosimeter, validated software for calculation of parameters, filters (1 mm or 2 mm Cu or 21 mm Al)
Procedure Use the image sequence acquired for the BNU test.
1) Linearize the image to air kerma using the response function.
2) Extract a centrally positioned ROI from the image. The ROI should cover most
of the FoV, i.e., leaving a gap of ~1 cm at the image edge.
3) From this large ROI, extract 2 x 2 mm half-overlapping small ROIs.
4) Within each small ROI, calculate the mean pixel value (SI ), variance (σ 2) and standard
Evaluation deviation (σ) of the pixel values.
5) Record the value of σ2 and SNR = SI/σ at the ROI position and repeat for all small ROIs
extracted from the large ROI, to generate the variance and SNR images.
6) Examine the variance image for signs of artefacts; regions with reduced variance that
could indicate blurring in the X-ray converter; regions with increased variance that
could indicate regions with increased electronic noise or artefacts.
The variance should change smoothly across the image with no abrupt changes and no
Pass/Fail Criteria
visible artefacts
Acceptance test
Frequency
Annual test

B.6.5 Noise Decomposition using Variance

To quantify the relative contribution of electronic (additive), quantum and

Aim
structure (multiplicative) noise sources to total noise (variance)
1. Monnin P, Bosmans H, Verdun FR, Marshall NW. Comparison of the polynomial model
against explicit measurements of noise components for different mammography systems.
Phys Med Biol. 2014;59(19)
2. Marshall NW, Kotre CJ, Robson KJ, Lecomber AR. Receptor dose in digital fluorography:
Reference A comparison between theory and practice. Phys Med Biol. 2001;46(4).
3. Mackenzie A, Honey ID. Characterization of noise sources for two generations of
computed radiography systems using powder and crystalline photostimulable
phosphors. Med Phys [Internet]. 2007;34(8):3345–57. Available from:
[Link]
Dosimeter, validated software for calculation of parameters, filters
Instrumentation
(1 mm or 2 mm Cu or 21 mm Al)
Use the image series acquired for the response function (preferably a fluoroscopy sequence
Procedure
as this is likely to have the largest contribution of electronic noise).

version 01.2024 49
 1) Measure the variance in a 5 x 5 mm ROI at the centre of the FoV.
2) Do this for all the series acquired i.e., all the IR-IAK/image levels.
3) Plot variance as a function of IR-IAK /image
4) Fit a 2nd order polynomial curve to the variance, weighting the variance data points by 1/
(IR-IAK/image)2 when performing the curve fit. In this model, total variance is the sum of
variance due to electronic, quantum and structure noise sources.

Evaluation 5) Record the fit coefficients for electronic (e), quantum (q) and structure noise (s)
(variance) – track these coefficients over the lifetime of the Image detector.
6) Using the fit coefficients, calculate the electronic, quantum and structure variance
at each IR-IAK level.
7) For each IR-IAK level, calculate the fraction of variance for each source (e, q and s)
compared to the total variance. Quantum noise should form the largest fraction of total
image noise at the standard clinical IRIAK/image for the imaging mode tested.
a. Electronic noise is dominant from an IR-IAK /image given by e/q
b. Structure noise is dominant from an IR-IAK /image given by q/s
8) Repeat the curve fitting for other imaging modes (fluoroscopy, cardiac imaging,
and series imaging) if tested.
There should be a simple relationship between variance and IR-IAK; variance should be
Pass/Fail Criteria described by the three noise components. At typical IR-IAK /image for each imaging mode,
quantum variance should form >50% of total variance.
Commissioning: Fluoroscopy, Cardiac or Series Modes (using the relevant response function)
Frequency
Annual: Fluoroscopy mode.

B.6.6 Noise Power Spectrum (NPS)

Aim To assess the magnitude of noise power as a function of spatial frequency

1. Granfors PR, Aufrichtig R, Possin GE, Giambattista BW, Huang ZS, Liu J, et al. Performance
of a 41×41 cm2 amorphous silicon flat panel x-ray detector designed for angiographic
and R&F imaging applications. Med Phys. 2003;30(10):2715–26.
Reference
2. Dobbins JT, Samei E, Ranger NT, Chen Y. Intercomparison of methods for image quality
characterization. II. Noise power spectrum). Med Phys [Internet]. 2006;33(5):1454.
Available from: [Link]
Instrumentation Dosimeter, validated software for calculation of parameters, filters (1 mm or 2 mm Cu or 21 mm Al)
1) The images obtained for the response function are used for the NPS calculation.
2) Select the IR-IAK /level for which the NPS will be calculated: this IR-IAK /image should
Procedure match the value used clinically for the imaging mode selected. This IR-IAK/image should
be used for NPS evaluation throughout the lifetime of the detector unless there is a large
change in the IR-IAK/image used clinically.

1) To calculate the NPS, select the image sequence acquired at the relevant IR-IAK level.
2) Extract a large region from the image centre (e.g. 576 x 576 from a 1024 x 1024 image).
3) It is typical in NPS algorithms to remove large area intensity trends that are present in
the region used to calculate the NPS, a process often referred to as ‘de-trending’. One
means of achieving this is to fit a 2D polynomial surface to this region and then subtract-
ing this 2D surface from the region
4) Extract NPS ROIs of 128 x 128 pixels from this large region in a half-overlapping pattern
and apply the NPS algorithm to these ROIs (Dobbins et al., 2006).
Evaluation 5) Once the 2D NPS has been calculated, obtain 1D sections of the NPS along the 0° and
90° Fourier axes, corresponding to the horizontal and vertical directions across the de-
tector. The NPS should be sectioned, 7 bins on each side of the axis, but the on-axis NPS
values should be excluded.
6) Normalise the NPS: divide by mean (linearized pixel value)² of the region used for the
NPS analysis. Most software will do this automatically or give this as an option.
7) Record the NNPS at 0.5 mm-1 and 2 mm-1 (or 1.5 mm-1 for a binned mode)
8) At the Commissioning stage consider assessing the NPS at the additional IR-IAK levels
i.e., 1/3.2 and 3.2 times the normal value for the Imaging Mode chosen.

50 QUALITY CONTROL IN DIGITAL BREAST TOMOSYNTHESIS (DBT) / EFOMP PROTOCOL

 Note: NNPS is likely to be subject to the influence of lag which will reduce NNPS compared
to a measurement made without the presence of lag. This lag can originate within the Image
Evaluation
detector (the X-ray converter layer and the pixels) or may also include some form of applied
post processing (frame averaging or recursive temporal filtering).
Record the normalised NPS (NNPS) at 0.5 mm-1 and 2 mm-1 (or 1.5 mm-1 for a binned
Pass/Fail Criteria mode). NNPS at these frequencies should be within ±15% of the baseline value
Annual test
Commissioning: Fluoroscopy, Cardiac or Series Modes (using the relevant response function)
Frequency
Annual test: Fluoroscopy mode.

B.7. IMAGE QUALITY

B.7.1 Signal-to-noise ratio rate, SNR2rate

Aim To estimate Signal-to-noise ratio1 rate SNR 2rate using a model observer
1. Tapiovaara, M. J. (1993) SNR and noise measurements for medical imaging: II. Applica-
tion to fluoroscopic x-ray equipment. Phys. Med. Biol, 38, pp. 1761–1788.
2. Tapiovaara, M. J. (2003) Objective measurement of image quality in fluoroscopic x-ray
equipment: FluoroQuality. STUK-A196 / MAY 2003.
3. Tapiovaara M.J. and Sandborg M How should low-contrast detail detectability be meas-
ured in fluoroscopy? Medical Physics 31(9) 2564-2576, (2004) doi: 10.1118/1.1779357
Reference 4. Tesselaar, E. and Sandborg, M. (2016) Assessing the usefulness of the quasi-ideal ob-
server for quality control in fluoroscopy. Radiation Protection Dosimetry, 169(1), pp.
360–364. doi: 10.1093/rpd/ncv434.
5. Elgström, H., Tesselaar, E. and Sandborg, M. (2021) Signal-to-noise ratio rate meas-
urements in fluoroscopy for quality control and teaching good radiological technique.
Radiation Protection Dosimetry, 195(3–4), pp. 407–415. doi: 10.1093/rpd/ncaa222.

Please note that the signal in B.7.1 is really a difference between the signal behind the test object and the signal in the back-
1 

ground beside the test object. It could therefore be denoted SDNR, but in order to keep the same notation as in references 1-5
we use SNR.

Patient simulating material

Setup 1: PMMA or equivalent slabs from 5 to 30 cm (at least 30 x 30 cm2 area)
Setup 2: Cu sheets of a range of thicknesses e.g., 0.5, 1, 2, 3 mm thick etc, large enough
to cover the whole x-ray beam at the collimator
TIQ test object
One or more small, well-defined target(s) are needed that can be added/removed from
the tissue simulating phantom without moving the phantom.
Instrumentation
Size: 1 to 10 mm effective diameter
Composition: e.g., aluminium, iron, iodine simulating material, a stent, bone simulating material.
Thickness: sufficient to generate a signal that is clearly visible over a range of patient phantom
thicknesses.
Software
Software (MATLAB-script) to compute SNR 2rate can be obtained (for non-commercial use)
from the EFOMP website. The software must not be used on patients as it is not a medical device.

version 01.2024 51
 TIQ using scattering phantom and ‘system’ geometry (setup 1)
1) Establish how many fluoro images the system can store from a single run (consult the
system manual). Approximately 1000 images are recommended. If only a smaller number
of images can be stored, then several sequences should be acquired and appended during
the analysis. Make sure not to include the first images that are used to stabilise the
acquisition.
2) Remove the mattress and align the patient simulating phantom with the central beam
axis, select the fluoroscopy mode or acquisition program to be tested, FoV,
source-image distance (SID), and table height.
3) Place the detail test object at the required position within the FoV e.g., on top of the
PMMA. See Figure 3b.
4) Perform an initial fluoroscopy sequence so the system selects the x-ray factors for the
TIQ phantom.
5) Record the air kerma-area product (PKA) before the image sequence.
6) Perform the imaging and store the image sequence. Record the acquisition parameters:
program/mode, pulse rate, tube voltage, tube current, pulse width, spectral pre-filter,
Procedure etc. or consult the DICOM header or structured report information.
7) Record the air kerma-area product (PKA) after the image sequence and the fluoroscopy
run time (t). Compute the P(t)KA by (P(t)KA – P(t)KA)/t.
8) Remove the detail test object, repeat the fluoroscopy sequence to acquire the signal ab-
sent images and store the image sequence. Ensure the system selects the same kV, mA,
added filtration etc. as used for the signal present sequence. This step is not required if
background ROIs are extracted next to the detail(s) in the signal present sequence.
9) Repeat for other imaging protocols and FoVs as needed.

TIQ using ‘detector’ geometry (setup 2)

1) Position the test object detail directly on the x-ray detector housing with the x-ray tube
positioned at the top (Figure 3a). This method is suited for mobile C-arm systems but
could also be used for fixed installations. Use 2 mm copper sheet mounted on the x-ray
beam collimator instead of PMMA to attenuate the beam to match that for a
normal-sized patient. Use 0.5 mm Cu for mini-C-arms.
2) Follow steps 4-9 from Setup 1 to compute the SNR 2rate for the ‘detector’ setup

Compute the SNR 2rate and associated uncertainty using the Matlab-analysis software. Typical
Evaluation
uncertainty is 4-7% (1 ± standard deviation) using 1024 images in a sequence.

SNR 2rate should be within ±20% of the baseline value for the acquisition program and dose
Pass/Fail Criteria
mode measured at Commissioning
Commissioning
Frequency
Annual test (one clinically relevant set-up)

B.7.2 Threshold-contrast detail detectability (TCDD) – Statistical method

Aim To estimate TCDD in the form of C TSM using the statistical method
1. Paruccini, N. et al. (2021) ‘A single phantom, a single statistical method for low-contrast
Reference detectability assessment’, Physica Medica, 91(October), pp. 28–42.
doi: 10.1016/[Link].2021.10.007
Patient simulating material
PMMA or equivalent slabs from 5 to 30cm (at least 30 x 30 cm2 area)
Cu sheets of a range of thicknesses e.g., 0.5, 1, 2, 3 mm thick etc and large enough
to cover the whole x-ray beam at the collimator.
TIQ test object
An aluminium plate of size 3 cm x 3 cm and thickness 0.5 ± 0.02 mm thick, of purity 99.9%.
Instrumentation
A step wedge with 5 steps, with thickness ranging from 0.25 to 1.25 mm in 0.25 ± 0.01 mm
steps, each step has x-y size 0.6 cm x 0.8 cm. The step wedge is used to convert contrasts
defined in pixel values to equivalent Aluminium (mm).
Software
Software can be obtained from [Link] for non-commercial use.
The software must not be used on patients as it is not a medical device.

52 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

 TIQ using scattering phantom and ‘system’ geometry
1) Remove the mattress and select the fluoroscopy mode or acquisition program
to be tested, FoV.
2) Place the patient simulating material on the table and the statistical method test object
on top of the phantom or on the entrance side (see Figure 3b).
3) Adjust table height so that the scattering phantom is at the system isocenter.
4) Adjust SID so there is a 5 cm gap between PMMA phantom and detector housing.
5) Once setup, perform an initial fluoroscopy sequence so the system selects the x-ray
factors for the TIQ phantom.
6) Perform fluoroscopy or series acquisition in order to obtain at least 30 images for the
TIQ analysis.
7) Save the fluoroscopy images.
8) Transfer the images and compute C TSM
Procedure
TIQ using ‘detector’ geometry
1) Position the test object detail directly on the x-ray detector housing with the x-ray tube
positioned at the top (Figure 3a). This method is suited for mobile C-arm systems but
could also be used for fixed installations.
2) Select the fluoroscopy mode or acquisition program to be tested, FoV, source-image
distance, and SID (this is normally fixed on a mobile C-arm).
3) Add copper sheets at the x-ray tube collimator until x-ray factors (kV, mA, ms, focus,
pre-filter etc) match those for a normal-sized patient.
4) Record the x-ray factors.
5) Perform fluoroscopy or series acquisition in order to obtain at least 30 images
for analysis. Save the fluoroscopy images.
6) Repeat other imaging protocols and FoVs as needed.
7) Transfer the images and compute C TSM
Analysis can be carried out with the dedicated software.
Alternatively, the following steps should be implemented.
1) Define a main ROI of 120 × 120 pixels within a large aluminium region.
2) Extract a series of smaller ROIs (sub-ROIs) with a binning size starting from 2x2 pixels.
3) For a given sub-ROI size (d × d), calculate the means of all the sub-ROIs (x), generate the
histogram of the mean values and take the standard deviation of this distribution (σx).
4) Calculate the threshold contrast for the statistical method (CTPV) in terms of pixel values,
defined using the 95% confidence level as:

CTPV (d)=3.29 σx (d)

Evaluation 5) Repeat for the range of sub-ROI sizes, e.g., from 2 × 2 to 24 × 24

6) Convert the raw threshold contrast values to Aluminium thickness equivalent. Measure
the mean pixel value in each step of the Al step wedge and plot against Al thickness. Ap-
ply a linear regression between mean pixel value and Al thickness for each image in the
series and record the slope (m). Calculate the threshold contrast for the statistical method
(C TSM ) in terms of Al (mm) as:

CTSM (d)=CTPV (d)*m

where m is the slope from the fit of MPV and Al (mm)

7) A
 minimum of 25 sub-ROIs should be extracted for each sub-ROI size to give a minimum
level of precision for the measurement.
Pass/Fail Criteria C TSM should be within ±20% of the baseline value for the imaging protocol /dose mode.
Commissioning
Frequency
Annual test

version 01.2024 53
B.7.3 Threshold-contrast detail detectability (TCDD) – CHO readout

Aim To estimate TCDD (C TCHO) using a CHO readout of the Leeds C-D test objects
1. Bertolini, M. et al. (2019) ‘Characterization of GE discovery IGS 740 angiography
system by means of channelized Hotelling observer (CHO)’, Physics in Medicine and
Biology. IOP Publishing, 64(9). doi: 10.1088/1361-6560/ab144c.
2. Favazza, C. P. et al. (2015) ‘Implementation of a channelized Hotelling observer model
to assess image quality of x-ray angiography systems’, Journal of Medical Imaging,
Reference
2(1), p. 015503. doi: 10.1117/[Link].2.1.015503.
3. Ortenzia, O. et al. (2020) ‘Radiation dose reduction and static image quality assessment
using a channelized hotelling observer on an angiography system upgraded with clarity
IQ’, Biomedical Physics and Engineering Express. IOP Publishing, 7(2).
doi: 10.1088/2057-1976/ab73f6.
Patient simulating material
PMMA or equivalent slabs from 5 to 30 cm (at least 30 x 30 cm2 area)
Cu sheets of a range of thicknesses e.g., 0.5, 1, 2, 3 mm thick and large enough to cover
the whole x-ray beam at the collimator.
TIQ test object
Instrumentation Use the available TO (e.g., Leeds TO10 or TO12 for fluoroscopy and angiography, or Leeds
TO16 or TO20 only for angiography acquisitions). Detail diameters from rows from 4.0 mm
to 0.35 mm diameter should be used for the analysis.
Software
Software (MATLAB-script) to compute the contrast detail curves can be obtained
(for non-commercial use) from the EFOMP website.

TIQ using scattering phantom and ‘system’ geometry

1) Approximately 400 images are required for the CHO calculation. If the system can only
store a smaller number of images, then acquisitions should be repeated and the images
added together in the analysis to improve the precision.
2) Remove the mattress and select the fluoroscopy mode or acquisition program to be
tested, FoV, SID.
3) Place the patient simulating material on the table and the Leeds test object e.g., on top
of the phantom (see Figure 3b). Adjust couch height so that the scattering phantom
is at the system isocentre.
4) Once setup, perform an initial fluoroscopy sequence so the system selects the x-ray
factors for the TIQ phantom.
5) Perform the fluoroscopy and store the image sequence.
6) Remove the test object from the beam and repeat the same fluoroscopy sequence for the
same length of time and store the image sequence for the signal absent images. Check
that the imaging system selects the same settings (kV, mA, added filtration etc.) as for
the sequence with test object, otherwise substitute the Leeds TO with a blank slab with
the same attenuation. This step can be omitted if the background ROIs are extracted from
Procedure
positions near the targets (discs) in the test object image.
7) Repeat for other imaging protocols and FoVs as needed.
8) Transfer the images and compute C TCHO

TIQ using ‘detector’ geometry

1) Position the Leeds test object detail directly on the x-ray detector housing with the x-ray
tube positioned at the top (Figure 3a). This method is suited for mobile C-arm systems but
could also be used for fixed installations.
2) Select the fluoroscopy mode or acquisition program to be tested, FoV, SID.
3) Add copper sheets at the x-ray tube collimator until x-ray factors (kV, mA, ms, focus,
pre-filter etc) match those for a normal-sized patient. Record the x-ray factors.
4) Once setup, perform an initial fluoroscopy sequence so the system selects the x-ray
factors for the TIQ phantom.
5) Perform fluoroscopy in order to obtain at least 400 images for analysis.
Save the fluoroscopy images.
6) Repeat for other imaging protocols and FoVs if needed.
7) Transfer the images and compute C TCHO

54 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Evaluation Analysis can be carried out with the dedicated software.
Pass/Fail Criteria C TCHO should be within ±20% of the baseline value for the imaging protocol/dose mode.
Commissioning
Frequency
Annual test

B.7.4 Threshold-contrast detail detectability (TCDD) – CHO readout

Aim To estimate (TCDD C TCHO) using a CHO readout of the Leeds C-d test objects
1. Cohen, G., McDaniel, D. L. and Wagner, L. K. (1984) ‘Analysis of variations in contrast-
detail experiments’, Medical Physics, 11(4), pp. 469–473;
2. Marshall, N. W. et al. (1992) ‘Analysis of variations in contrast-detail measurements
performed on image intensifier-television systems’, Physics in Medicine and Biology,
37(12), pp. 2297–2302.
3. Launders, J. H. et al. (1995) ‘Update on the recommended viewing protocol for FAXIL
Reference threshold contrast detail detectability test objects used in television fluoroscopy’,
British Journal of Radiology, 68(805), pp. 70–77.
4. Tapiovaara, M. J. (1997) ‘Efficiency of low-contrast detail detectability in fluoroscopic
imaging.’, Medical physics, 24(5), pp. 655–64. Available at:
[Link]
5. Tapiovaara, M. J. and Sandborg, M. (2004) ‘How should low-contrast detail detectability
be measured in fluoroscopy?’, Medical Physics, 31(9), p. 2564. doi: 10.1118/1.1779357.
Patient simulating material
PMMA or equivalent slabs from 5 to 30cm (at least 30 x 30 cm2 area)
Cu sheets of a range of thicknesses e.g., 0.5, 1, 2, 3 mm thick etc and large enough to cover
Instrumentation the whole x-ray beam at the collimator.

TIQ test object

Use a suitable contrast detail test object (e.g., Leeds TO10 for fluoroscopy and TO12,
TO16 or TO20 for fluoroscopy and angiography).
TIQ using scattering phantom and ‘system’ geometry
1) Remove the mattress and select the fluoroscopy mode or acquisition program to be test-
ed, FoV, SID.
2) Place the patient simulating material on the table and the Leeds test object e.g., on top
of the phantom or at the entrance side (see Figure 3b). Adjust couch height so that the
scattering phantom is at the system isocenter.
3) Perform an initial fluoroscopy sequence so the system selects x-ray factors.
4) Perform the imaging fluoroscopy
5) Use a consistent viewing protocol such that the number of discs seen is maximised and
visually score the images.
6) The result can be left as a ‘number of discs’.
7) Repeat for other imaging protocols and FoVs as needed

Procedure TIQ using ‘detector’ geometry

1) Position the Leeds test object detail directly on the x-ray detector housing with the x-ray
tube positioned at the top (Figure 3a). This method is suited for mobile C-arm systems but
could also be used for fixed installations.
2) Select the fluoroscopy mode or acquisition program to be tested, FoV, distance and SID.
3) Add copper sheets at the x-ray tube collimator until x-ray factors (kV, mA, ms, focus,
pre-filter etc) match those for a normal-sized patient. Record the x-ray factors.
4) Once setup, perform an initial fluoroscopy sequence so the system selects the x-ray
factors for the TIQ phantom.
5) Perform fluoroscopy
6) Use a consistent viewing protocol such that the number of discs seen is maximised and
visually score the images.
8) The result can be left as a ‘number of discs’.
7) Repeat for other imaging protocols and FoVs as needed.
Evaluation Record the number of details and compare with previous evaluations.
Pass/Fail Criteria NA
Commissioning
Frequency
Annual test

version 01.2024 55
B.7.5 Limiting spatial resolution (LSR)

Aim To estimate limiting spatial resolution with a line pair test object
1. Albert, M. et al. (2002) ‘Aliasing effects in digital images of line-pair phantoms’, (August),
Reference
pp. 1716–1718. doi: 10.1118/1.1493212.
Patient simulating material
PMMA or equivalent slabs from 5 to 30cm (at least 30 x 30 cm2 area).
Instrumentation
LSR test object
A suitable line pair test object
LSR using scattering phantom and ‘system’ geometry
1) Remove the mattress
2) Select the most commonly used fluoroscopy mode, then select the relevant PMMA thickness
and FoV then set the SID.
3) Place the patient simulating material on the table and the line pair test object on top of
the phantom, in the centre or at the entrance side (see Figure 3b). Adjust couch height so
that the scattering phantom is at the system isocentre.
4) The line pair patterns should be positioned at approximately 45° to the pixel matrix.
5) Perform an initial fluoroscopy sequence so the system can select the relevant x-ray factors.
6) Perform the fluoroscopy imaging, fix the viewing conditions, and visually score the images
7) Repeat for a number of PMMA thicknesses, fluoroscopy modes, and FoVs.
Procedure
LSR using ‘detector’ geometry
1) Position the line pair test object detail directly on the x-ray detector housing with the x-ray
tube positioned at the top (Figure 3a). The bar patterns should be positioned at an angle
of approximately 45° to the pixel matrix. This method is suited for mobile C-arm systems
but could also be used for fixed installations. Do not place additional Cu filtration at the
x-ray tube.
2) Select the largest FoV and the fluoroscopy mode or acquisition program to be tested
3) Perform fluoroscopy, fix the viewing conditions, and visually score the test object images.
4) Repeat for the other FoVs.
5) Calculate the limiting spatial resolution
Evaluation Visual scoring
For ‘detector’ geometry:
Pass/Fail Criteria - the LSR should not differ from the baseline values by more than 25%
- the LSR should be compared against national regulations where they exist
Acceptance test (detector geometry)
Frequency Commissioning (system geometry and detector geometry)
Annual test (system geometry for FPD, system geometry and detector geometry for XRII)

56 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

SECTION C – REFERENCES

Abbey, C. K. and Barrett, H. H. (2001) ‘Human- and model-observer performance in ramp-spectrum noise : ef-
fects of regularization and object variability’, Journal of the Optical Society of America A, 18(3), pp. 473–488.

Albert, M. et al. (2002) ‘Aliasing effects in digital images of line-pair phantoms’, (August), pp. 1716–1718. doi:
10.1118/1.1493212.

American College of Radiology. ACR-AAPM (2016). ‘Technical Standard for Diagnostic Medical Physics Per-
formance Monitoring of Fluoroscopic Equipment’. Available at: [Link]
Practice-Parameters/[Link]?la=en. (Accessed 25 March 2020)

Andersson, J. et al. (2021) ‘Estimation of patient skin dose in fluoroscopy: summary of a joint report by AAPM
TG357 and EFOMP’, Medical Physics, 48(7), pp. e671–e696. doi: 10.1002/mp.14910.

Barrett, H. H. et al. (1993) ‘Model observers for assessment of image quality.’, Proceedings of the National
Academy of Sciences of the United States of America, 90(21), pp. 9758–65. Available at: [Link]
[Link]/[Link]?artid=47653&tool=pmcentrez&rendertype=abstract.

Barrett, H. H. and Myers, K. J. (2003) Foundations of Image Science. Wiley.

Båth, M. and Månsson, L. G. (2007) ‘Visual grading characteristics (VGC) analysis: a non-parametric rank-in-
variant statistical method for image quality evaluation.’, The British journal of radiology, 80(951), pp. 169–76.
doi: 10.1259/bjr/35012658.

Bednarek DR, Barbarits J, Rana VK, Nagaraja SP, Josan MS, Rudin S. Verification of the performance accuracy
of a real-time skin-dose tracking system for interventional fluoroscopic procedures. Proc SPIE–Int Soc Opt
Eng 2011;7961: 796127_1.

Bertolini, M. et al. (2019) ‘Characterization of GE discovery IGS 740 angiography system by means of chan-
nelized Hotelling observer (CHO)’, Physics in Medicine and Biology. IOP Publishing, 64(9). doi: 10.1088/1361-
6560/ab144c.

Boone, J. . et al. (1993) ‘A survey of fluoroscopic exposure rates: AAPM Task Group No. 11 Report’, Medical
Physics, 20(3), pp. 789–94.

Bordier C, Klausz R, Desponds L. Patient dose map indications on interventional X- ray systems and validation
with Gafchromic XR-RV3 film. Rad Prot Dosim 2015; 163:306–18.

Burgess, A. E. (2011) ‘Visual perception studies and observer models in medical imaging’, Seminars in Nuclear
Medicine. Elsevier Inc., 41(6), pp. 419–436. doi: 10.1053/[Link].2011.06.005.

Carmichael, J. H. E. et al. (1996) European guidelines on quality criteria for diagnostic radiographic images
EUR 16260 EN.

Chao, E. H. et al. (2000) ‘A Statistical Method of Defining Low Contrast Detectability. In: Oak Brook, editor.
Radiol. Soc. North Am., 2000.’, Radiology.

version 01.2024 57
Cohen, G., McDaniel, D. L. and Wagner, L. K. (1984) ‘Analysis of variations in contrast-detail experiments’,
Medical Physics, 11(4), pp. 469–473.

Cowen, A. R., Haywood, J. M., et al. (1987) ‘A set of X-ray test objects for image quality control in digital
subtraction fluorography. I: Design considerations.’ The British Journal of Radiology, 60(718), pp. 1001–9.

Cowen, A. R., Workman, A., et al. (1987) ‘A set of X-ray test objects for image quality control in digital sub-
traction fluorography. II: Application and interpretation of results.’ The British Journal of Radiology, 60(718),
pp. 1011–8.

Cunningham IA. Applied Linear-Systems Theory. In: Jacob Beutel, editor. Medical Imaging Volume 1 Physics
and Psychophysics. Bellingham: SPIE - The International Society for Optical Engineering; 2000. p. 79–156.

Dabin J et. al. Accuracy of skin dose mapping in interventional cardiology: Comparison of 10 software prod-
ucts following a common protocol. Physica Medica 2021 82 279-294

Dance, D. R. et al. (2014) Diagnostic Radiology Physics: A Handbook for Teachers and Students, Diagnostic
Radiology Physics: A Handbook for Teachers and Students. Vienna, Austria: International Atomic Energy
Authority (IAEA).

Doerfler, A. et al. (2015) ‘Flat-Panel Computed Tomography (DYNA-CT) in Neuroradiology. From High-Reso-
lution Imaging of Implants to One-Stop-Shopping for Acute Stroke’, Clinical Neuroradiology, 25, pp. 291–297.
doi: 10.1007/s00062-015-0423-x.

Dobbins III, James T., Ergun, David L., Rutz, Lois, Hinshaw, Dean A., Blume, Hartwig, Clark DC. DQE(f) of four
generations of computed radiography acquisition devices. Med Phys. 1995;22(10):1581–93.

Dobbins JT, Samei E, Ranger NT, Chen Y. Intercomparison of methods for image quality characteriza-
tion. II. Noise power spectrum). Med Phys [Internet]. 2006; 33(5):1454. Available from: [Link]
com/10.1118/1.2188819

Eckstein, M. P., Abbey, C. K. and Bochud, F. O. (2000) ‘A Practical Guide to Model Observers for Visual Detec-
tion in Synthetic and Natural Noisy Images’, in Beutel, J., Kundel, H. L., and Van Metter, R. L. (eds) Handbook
of Medical Imaging, Volume 1. Physics and Psychophysics. SPIE, Bellingham, WA, pp. 595–628.

Elgström, H., Tesselaar, E. and Sandborg, M. (2021) ‘Signal-to-noise ratio rate measurements in fluoroscopy
for quality control and teaching good radiological technique’, Radiation Protection Dosimetry, 195(3–4), pp.
407–415. doi: 10.1093/rpd/ncaa222.

European Commission (2012) RP 162. Criteria for Acceptability of medical radiological equipment used in
diagnostic radiology, nuclear medicine and radiotherapy, Radiation Protection 162. doi: 10.2768/22561.

Fahrig, R. et al. (2021) ‘Flat-panel cone-beam CT in the clinic: history and current state’, Journal of Medical
Imaging, 8(05), pp. 1–40. doi: 10.1117/[Link].8.5.052115.

Favazza, C. P. et al. (2015) ‘Implementation of a channelized Hotelling observer model to assess image quality
of x-ray angiography systems’, Journal of Medical Imaging, 2(1), p. 015503. doi: 10.1117/[Link].2.1.015503.

58 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Gallas, B. D. and Barrett, H. H. (2003) ‘Validating the use of channels to estimate the ideal linear observer.’
Journal of the Optical Society of America. A, Optics, image science, and vision, 20(9), pp. 1725–38. Available
at: [Link]

Gardavaud F, Tavolaro S, Gussenmeyer-Mary N, Cornelis F, Boudgh`ene F. PeakSkin Dose evaluation for vas-
cular clinical procedures in interventional radiology: a comparison between three computation numerical
solutions. Phys Med 2018; 56

Granfors, P. R. et al. (2003) ‘Performance of a 41×41 cm2 amorphous silicon flat panel x-ray detector designed
for angiographic and R&F imaging applications’, Medical Physics, 30(10), pp. 2715–2726.

Green, D. M. and Swets, J. A. (1966) Signal detection theory and psychophysics. John Wiley.

Habib Geryes B, Hadid-Beurrier L, Waryn MJ, Jean-Pierre A, Farah J. Benchmarking the DACS-integrated
Radiation Dose Monitor (R) skin dose mapping software using XR-RV3 Gafchromic (R) films. Med Phys 2018;
45:4683–92.

Hay, G. A. et al. (1985) ‘A set of X-ray test objects for quality control in television fluoroscopy.’ The British
Journal of Radiology, 58(688), pp. 335–344.

He, X. and Park, S. (2013) ‘Model Observers in Medical Imaging Research’, Theranostics, 3(10), pp. 774–786.
doi: 10.7150/thno.5138.

de las Heras Gala, H. et al. (2017) ‘Quality control in cone-beam computed tomography (CBCT) EFOMP-ES-
TRO-IAEA protocol (summary report)’, Physica Medica, 39, pp. 67–72. doi: 10.1016/[Link].2017.05.069.

Hiles, P. and Starritt, H. (1996) IPEM Report 32 (II) Measurement of the performance characteristics of diag-
nostic x-ray systems used in medicine: x-ray image intensifier television systems. York, UK.

Holm, T. and Moseley, R. D. (1964) ‘The Conversion Factor for Image Intensifiers’, Radiology, 82(January), pp.
898–904. doi: 10.1148/82.5.898.

IEC 62220-1-3 (2008) Medical electrical equipment - Characteristics of digital X-ray imaging devices - Part 1-3:
Determination of the detective quantum efficiency - Detectors used in dynamic imaging. Geneva, Switzerland.

IAEA. Dosimetry in diagnostic radiology: an international code of practice. — Vienna: International Atomic
Energy Agency, 2007. p.; 24 cm. — (Technical reports series, ISSN 0074–1914; no. 457) STI/DOC/-10/457 ISBN
92–0–115406–2.

[Link]

International Atomic Energy Agency (2018) ‘Radiation Protection and Safety in Medical Uses of Ionizing
Radiation’, IAEA Safety Standards Series No. SSG-46, IAEA, Vienna. Available at: [Link]
lications/11102/radiation-protection-and-safety-in-medical-uses-of-ionizing-radiation

ICRU Rep. No. 54.’ (1996). Bethesda, MD.

IEC 60601-2-43:2022. Medical electrical equipment - Part 2-43: Particular requirements for the basic safety
and essential performance of X-ray equipment for interventional procedures

version 01.2024 59
IEC 60601-2-54: 2022 (. Medical electrical equipment - Part 2-54: ‘Particular requirements for the basic safety
and essential performance of X-ray equipment for radiography and radioscopy.

IEC 62220-1-1: 2015. Medical electrical equipment - Characteristics of digital X-ray imaging devices - Part 1-1:
Determination of the detective quantum efficiency - Detectors used in radiographic imaging.

IEC 62220-1-2: 2007. Medical electrical equipment - Characteristics of digital X-ray imaging devices - Part 1-1:
Determination of the detective quantum efficiency - Detectors used in mammography.

IEC 62220-1-3: 2008. Medical electrical equipment - Characteristics of digital X-ray imaging devices - Part 1-3:
Determination of the detective quantum efficiency - Detectors used in dynamic imaging.

IPEM (2005) Recommended standards for the performance testing of diagnostic x-ray imaging systems,
IPEM Report No. 91. York, UK.

Jones AK & Pasciak AS 2011, Calculating the peak skin dose resulting from fluoroscopically guided interven-
tions. Part I: Methods, Journal of Applied Clinical Physics 12 402

Jones, A. K. et al. (2014) ‘Medical imaging using ionizing radiation: Optimization of dose and image quality in
fluoroscopy’, Medical Physics, 41(1), pp. 014301-1–26. doi: 10.1118/1.4835495.

Kyprianou IS, Rudin S, Bednarek DR, Hoffmann KR. Generalizing the MTF and DQE to include x-ray scatter
and focal spot unsharpness: Application to a new microangiographic system. Med Phys. 2005;32(2):613.

Launders, J. H. et al. (1995) ‘Update on the recommended viewing protocol for FAXIL threshold contrast de-
tail detectability test objects used in television fluoroscopy’, British Journal of Radiology, 68(805), pp. 70–77.

Lin, P. J. P. et al. (2022) ‘AAPM Task Group Report 272: Comprehensive acceptance testing and evaluation of
fluoroscopy imaging systems’, Medical Physics, 49(4), pp. e1–e49. doi: 10.1002/mp.15429.

Lin, P. J. P., Goode, A. R. and Corwin, F. D. (2022) ‘Review and investigation of automatic brightness/dose
rate control logic of fluoroscopic imaging systems in cardiovascular interventional angiography’, Radiological
Physics and Technology. Springer Singapore, 15(1), pp. 6–24. doi: 10.1007/s12194-022-00649-3.

Lundh, C. et al. (2021) ‘A model for evaluating the use of imaging in image-guided interventional procedures
- Possible implications on optimisation of radiation protection’, Radiation Protection Dosimetry, 195(3–4), pp.
139–144. doi: 10.1093/rpd/ncab040.

Mackenzie, A. et al. (2010) IPEM Report 32 (Part VII) Measurement of the performance characteristics of
diagnostic x-ray systems: digital imaging systems. York, UK.

Malchair F et. al. 2020 Review of skin dose calculation software in interventional cardiology. Physica Medica
80 75-83

Mackenzie A, Doshi S, Doyle P, Hill A, Honey I, Marshall NW, et al. IPEM Report 32 (Part VII) Measurement of
the performance characteristics of diagnostic x-ray systems: digital imaging systems. York, UK; 2010.
Mackenzie A, Honey ID. Characterization of noise sources for two generations of computed radiography
systems using powder and crystalline photostimulable phosphors. Med Phys [Internet]. 2007;34(8):3345–57.
Available from: [Link]

60 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Malusek A, Helmrot E, Sandborg M, Grindborg J-E and Alm Carlsson G In-situ calibration of clinical built-
in KAP meters with traceability to a primary standard using a reference KAP-meter. Phys. Med. Biol. 59,
7195–7210 (2014), doi:10.1088/0031-9155/59/23/7195

Marshall, N. W. et al. (1992) ‘Analysis of variations in contrast-detail measurements performed on image

intensifier-television systems’, Physics in Medicine and Biology, 37(12), pp. 2297–2302.

Marshall, N. W. et al. (2001) ‘Receptor dose in digital fluorography: A comparison between theory and prac-
tice’, Physics in Medicine and Biology, 46(4). doi: 10.1088/0031-9155/46/4/325.

Marshall NW, Mackenzie A, Honey ID. Quality control measurements for digital x-ray detectors. Phys Med Biol.
2011;56(4).

Menzel, H.-G., Schibilla, H. and Teunen, D. (1999) European Guidelines on Quality Criteria for Computed To-
mography European Guidelines on Quality Criteria, Eur 16262 En.

Metz, C. E. (1986) ‘ROC Methodology in Radiologic Imaging’, Investigative Radiology, 21, pp. 720–733.

Monnin P, Bosmans H, Verdun FR, Marshall NW. Comparison of the polynomial model against explicit mea-
surements of noise components for different mammography systems. Phys Med Biol. 2014;59(19).

National Council on Radiation Protection and Measurements (2010) ‘Radiation Dose Management for Fluo-
roscopically-Guided Interventional Medical Procedures, NCRP Report No. 168’. National Council on Radiation
Protection and Measurements, Bethesda, Maryland.

NEMA (2018) ‘NEMA Standards Publication XR 27-2013 (R2018) X-ray Equipment for Interventional Proce-
dures User Quality Control Mode’. National Electrical Manufacturers Association (NEMA), Virginia, USA.

Official Journal of the European Union (2014) ‘European Council Directive 2013/59/Euratom on basic
safety standards for protection against the dangers arising from exposure to ionising radiation and re-
pealing directives 89/618/Euratom, 90/641/Euratom, 96/29/Euratom, 97/43 Euratom and 2003/122/
Euratom’. OJ of the EU. L13;57:1-73 Available at: [Link]
J:L:2014:013:0001:0073:EN:PDF . (Accessed 14 Mar 2020)

Paruccini, N. et al. (2021) ‘A single phantom, a single statistical method for low-contrast detectability assess-
ment’, Physica Medica, 91(October), pp. 28–42. doi: 10.1016/[Link].2021.10.007.

Rose, A. (1948) ‘The sensitivity performance of the human eye on an absolute scale’, J. Opt. Soc. Am., 38, pp. 196–208.

Samei E, Ranger NT, Dobbins JT, Chen Y. Intercomparison of methods for image quality characterization. I.
Modulation transfer function. Med Phys. 2006;33(5):1454–65

Shepard, S. J. (2002) Quality Control in Diagnostic Radiology AAPM Report 72, Report of Task Group #12
Diagnostic X-ray Imaging Committee.

Smedby, O. and Fredrikson, M. (2010) ‘Visual grading regression: analysing data from visual grading experi-
ments with regression models.’, The British journal of radiology, pp. 1–9. doi: 10.1259/bjr/35254923.

Stevens, G. (2021) Dynamic X-ray Imaging Systems Used in Medicine. York, UK.

version 01.2024 61
Tapiovaara, J. (1993) ‘SNR and noise measurements for medical imaging: II. Application to fluoroscopic x-ray equip-
ment’, Phys. Med. Biol, 38, pp. 1761–1788. [Link]

Tapiovaara, M. (2003) Objective measurement of image quality in fluoroscopic x-ray equipment: FluoroQuality,
STUK-A196 / MAY 2003. Tapiovaara, M. J. (1997) ‘Efficiency of low-contrast detail detectability in fluoroscopic
imaging.’, Medical physics, 24(5), pp. 655–64. Available at: [Link]

Tapiovaara, M. J. and Sandborg, M. (2004) ‘How should low-contrast detail detectability be measured in
fluoroscopy?’, Medical Physics, 31(9), p. 2564. doi: 10.1118/1.1779357.

Tapiovaara, M. J. and Wagner, R. F. (1993) ‘SNR and noise measurements for medical imaging: I. A practi-
cal approach based on statistical decision theory’, Physics in Medicine and Biology, 38(1), pp. 71–92. doi:
10.1088/0031-9155/38/1/006.

Tesselaar, E. and Sandborg, M. (2016) ‘Assessing the usefulness of the quasi-ideal observer for quality control
in fluoroscopy’, Radiation Protection Dosimetry, 169(1), pp. 360–364. doi: 10.1093/rpd/ncv434.

Toroi P, Komppa T and Kosunen A 2008 A tandem calibration method for kerma–area product meters. Phys.
Med. Biol. 53 4941–58, DOI 10.1088/0031-9155/53/18/006

Vedantham, S. et al. (2004) ‘Solid-state fluoroscopic imager for high-resolution angiography : Physical char-
acteristics of an 8 cm × 8 cm experimental prototype Solid-state fluoroscopic imager for high-resolution
angiography : Physical characteristics of an 8 cm à 8 cm experimental p’, 1462. doi: 10.1118/1.1750992.

Villa, R. et al. (2019) ‘Model observers for Low Contrast Detectability evaluation in dynamic angiography: A
feasible approach’, Physica Medica, 64(June), pp. 89–97. doi: 10.1016/[Link].2019.06.015.

Worrall, M. et al. (2020) ‘IPEM Topical Report: An evidence and risk assessment based analysis of the efficacy
of quality assurance tests on fluoroscopy units - Part II; Image quality’, Physics in Medicine and Biology. IOP Publishing,
65(22), p. 195011. doi: 10.1088/1361-6560/abb92.

62 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

SECTION D – APPENDICES

D.1. Appendix 1. Radiation Safety Procedures

The performance evaluation of fluoroscopic systems should include a radiation safety survey to determine
whether the radiological medical equipment, the practical techniques and the ancillary equipment are com-
plying with the national radiation control regulations, based on the European Basic Safety Standards (BSS).
A summarised list of BSS requirements for fluoroscopes and interventional radiology equipment is shown in
the table below.

Summary of BSS Requirements applicable to Fluoroscopy/Angiography Systems

All fluoroscopy equipment has a device that automatically controls the dose rate, using an image inten-
sifier or equivalent device (flat panel detector) {Art. 60. 3. (a)}.

All equipment used for interventional radiology has a device or software that informs the operators at
the end of the procedure, of relevant parameters for assessing the patient dose. In addition, equipment
installed after 6 February 2018 has the capacity to transfer this information to the examination record.
{Art. 60. 3. (a)}

Interventional radiology equipment installed after 6 February 2018 has also a device or a feature inform-
ing the practitioner of the quantity of radiation produced by the equipment during the procedure. {Art.
60. 3 (d)}

Image acquisition protocols, at least for the most common interventional procedures, will be reviewed
to assess if “patient doses from interventional radiology are kept as low as reasonably achievable
consistent with obtaining the required medical information, taking into account economic and societal
factors”. {Art. 56. 1.}

Diagnostic reference levels have been established and are used for adult and paediatric interventional
radiology procedures. {Art. 56. 2.}

A medical physics expert is involved in all dosimetry {Art 58 (d) (ii) & Art 61. 1. (c)} and radiation safety
assessments, liaising with the radiation protection expert as appropriate {Art 83}.

Medical physicists can replace this list with the corresponding requirements in their national regulations.

The radiation safety survey can consist of a review of image acquisition protocols, staff radiation doses and
any relevant documentation such as previous safety reports; a visual inspection to check for equipment
integrity and ease of motions; verification of interlock functions, such as collision alerts, and stray (leakage
plus scatter) radiation measurements around the fluoroscope using the maximum air kerma rate and the
largest field size. The extent and depth of the survey will depend on whether it is performed at acceptance
testing and commissioning or during a periodic equipment performance evaluation.

version 01.2024 63
Stray radiation measurements

References See Section A.6

Procedure 1:
- PMMA 30x30x40 cm3 slab phantom.
Instrumentation - C alibrated survey meter or large volume ionisation chamber and electrometer with kerma
rate and integrated modes.
- Stand to hold the ion chamber in air.
Procedure 1 (Hand-held survey meter or ion chamber on a stand)
Set up the phantom perpendicular to the x-ray beam and position the measurement device at
various distances around the phantom, in the locations where the interventionalist(s) stand(s)
and where the auxiliary staff monitor the procedure’s technical parameters and the patient’s
vital signs. Typical measurement distances are 50 cm away from the centre of the phantom
and 20 cm above and below the table at 20 cm from the edge of the phantom. Turn on the
fluoroscope using the maximum field size. Record the settings given by the AEC (kV, mA, pulse
width, filtration) and the values of air kerma rate or ambient dose equivalent (depending on
how the instrument is calibrated) which are read by the survey meter or ion chamber in the
different positions.

Figure Appendix 1-1, below, shows a diagram of a simple set-up.

Procedure

Assuming that the leakage component of the secondary radiation is negligible compared
with the scatter radiation, the measured Air Kerma Rates may be compared with the iso-scat-
ter plots found in the Manufacturer’s IUF/Accompanying Documents, by normalising to the
same technique settings. An illustration of such plots is shown in Figure Appendix 1-2. Below.

Evaluation

Pass/Fail Criteria The measurement conditions and the resulting values should be recorded for future tests.
At acceptance testing and when a major system component such as the x-ray tube has been
Frequency
replaced.

64 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

D.2. Appendix 2. Phantoms

The most common means currently of evaluating technical image quality in QC testing is with test objects,
where the parameters assessed are usually low-contrast detectability and limiting spatial resolution.

Threshold-contrast detail detectability (TCDD) test objects

Threshold low contrast detail detectability (TCDD) is assessed using threshold contrast-detail test objects,
which have a long history in the evaluation of imaging devices, including screen/film, XRII and TV camera
tubes since the 1950’s (Rose, 1948; Burger, 1950; Hay et al., 1985; Cowen, Workman, et al., 1987; Thijssen, 1993;
Karssemeijer and Thiijssen, 1996). These test objects are usually arranged in a homogeneous background
and generate a set of stimuli with varying diameter and contrast. For each detail diameter, the visibility falls
systematically until the disc is no longer visible. The job of the observer is to determine the last stimulus con-
sidered visible for each diameter and the contrast of this stimulus is defined as the threshold contrast (CT).
The stimulus for which this occurs depends on the signal to noise ratio (SNR) in the image and the system
parameters that influence this will depend on how the test object is imaged. For example, if the test objects
are placed at the detector input plane (~’detector’ geometry), at some prescribed x-ray energy then the most
influential parameters will be x-ray detector exposure rate used and the DQE of the x-ray detector (Cowen,
Haywood, et al., 1987). TCDD results can therefore be used as an overall measure of technical imaging perfor-
mance, as quantified by the SNR or SDNR, of either the imaging system or the x-ray detector, depending on
how the TCDD test object is imaged.

Figure Appendix 2-1 a) image of Leeds TO12 TCDD test object, an example of “count to the last visible disc” design; b) CDRAD
test object, an example of an alternative forced choice design (4-AFC here).

TCDD types, metric for the test object score and test object composition
TCDD test objects generally have a detail plate, typically made of PMMA or Aluminium, that contains the
details used to generate the stimuli. These plates can be imaged with a stack of (typical) patient simulating
material, so that the system selects x-ray imaging parameters relevant to patient imaging. This is usually
PMMA blocks but a few millimetres of Cu or Al can also be used. Many test objects follow the initial design
described by Albert Rose (Rose, 1948) and generate circular signals of varying diameter against a homoge-
neous background, using either discs of varying thickness or holes of different depths drilled into the plate.
When imaged, this generates a set of signals whose contrast and therefore visibility falls until the disc is no
longer visible. As discussed above the arrangement of the discs across will be different depending on the
readout method envisioned. Some test objects, for example CDRAD (Thijssen, 1993) have a disc randomly
positioned in a corner of each square together with a disc at the square, enabling the use of both multiple

version 01.2024 65
alternative forced choice (MAFC) and ‘count until the last disc’ readout methods. A range of different TCDD
test objects are available from different suppliers. Details of test object design and composition, where avail-
able, are given in Table Appendix 2-1.

It is important to consider the output value produced by the test object, as this will be used to express the
image quality score. For test objects that use holes of different depths in PMMA or Aluminium to generate
signals, such as CDRAD, CDDISC or the Gammex model 151, the image quality score can be given as hole
depth (mm). Expressed like this, then the image quality score is effectively test object specific and comparing
between systems requires the same test object to be used. This is useful for longitudinal testing against a
baseline but cannot be used to set a typical performance standard for a common procedure unless physicists
all use the same test object. It might be possible to convert this to a threshold contrast for a given beam
quality via calibration measurements (Aufrichtig, 1999) but this is difficult to achieve for fluoroscopy, given
the range of possible factors that can be selected by these systems. A further difficulty is scatter adiation,
where the scatter fraction is very sensitive to the geometry.

An alternative approach is to use a detail plate in a low scatter geometry, as done with the Leeds test ob-
jects. The specific materials used for discs in these test objects are not specified. Instead, tables of radiation
contrasts are supplied with the test objects for tube voltages between 65 kV and 80 kV and added copper
filtration (1.0, 1.5 and 2.0 mm copper). Successive diameters in the Leeds test objects change by a factor of √2,
while for a given diameter there is a factor of √2 change in contrast between successive discs. For a quantum
noise limited system, a factor of 2 increase in detector exposure should visualise one extra disc at each
diameter (Rose, 1948). These objects are calibrated to produce a defined range of contrasts under low scatter
beam qualities at specific energies (Hay et al., 1985). By accounting for the energy used in the measurement
and imaging under low scatter conditions, an absolute limiting threshold contrast for the x-ray detector can
be estimated. If a typical performance level can be established for an imaging mode at some x-ray detector
input exposure rate, then reference curves can be formed and used for comparison (Evans, 2004). However,
used like this, the influence of spectral selection of the imaging system and anti scatter rejection technique
on target detectability is not assessed.

A further consideration regarding test object composition is the recent introduction of contrast to noise
ratio (CNR) imaging modes on fluoroscopy and angiography systems (Dehairs, Bosmans and Marshall, 2019;
Werncke et al., 2021a, 2021b). In these modes, the elemental composition of the target can be specified
and based on this information, the AEC selects x-ray factors that maximise the visibility of objects with this
composition. Examples can be iodine, carbon dioxide or barium used as contrast agent materials, iron used in
guidewires, and platinum or tantalum used in stent markers (Dehairs, Bosmans and Marshall, 2019; Werncke
et al., 2021b). Evaluation of material specific imaging modes with test objects containing targets of a different
elemental composition may give inconsistent or unexpected results and may have reduced relevance when
trying to predict system performance when imaging patients (Lin, Goode and Corwin, 2022). Some test
objects are available with specific elemental contrasts, for example NEMA XR21-2000 detail plate contains
cylindrical holes filled with different areal densities of iodine to generate contrast targets (Balter et al., 2001).

66 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Table Appendix 2-1. Some of the TCCD test objects available

Base plate Detail Detail

Analysis Image/
Name Design thickness/ # Details Diameters thicknesses/
software? diagram
composition (mm) hole depth
Cylindrical
Cylindrical holes:
holes: 0.3, 0.5,
0.3, 0.5, 0.6, 0.8,
MAFC/ 0.6, 0.8, 1.0,
1.0, 1.3, 1.6, 2.0,
Artinis Count 10 mm 1.3, 1.6, 2.0,
240 2.5, 3.2, 4.0 5.0, No
CDDISC to last PMMA 2.5, 3.2, 4.0
6.3, 8.0
disc 5.0, 6.3, 8.0
In exponential
In exponential
steps
steps

Cylindrical
Cylindrical holes: Artinis CD
MAFC/ holes: 0.3, 0.5,
0.3, 0.5, 0.6, 0.8, Analyser;
Artinis Count 10 mm 0.6, 0.8, 1.0,
225 1.0, 1.3, 1.6, 2.0, calculation
CDRAD to last PMMA 1.3, 1.6, 2.0,
2.5, 3.2, 4.0 5.0, method
disc 2.5, 3.2, 4.0
6.3, 8.0 not specified
5.0, 6.3, 8.0

CIRS Ra-
Aluminium
diography Count
layer (30 x Discs, 5.6, 4.0, 6 depths,
Fluoroscopy to last 24 No
70 x 9.53 2.8, 2.0 0.38 to 1.52 mm
QA Phantom disc
mm)
Model 903

Gammex LC
Count Al block,(L x
Resolution 0.58 to
to last W x H) 180 x 100 0.13 to 2.29 mm No
QC Tool 7.93 mm
disc 180 x 13 mm
(model 151)

14 thicknesses,
Discs, 11.1, 7.9, thickness and
AutoPIA
Count 10 mm thick 5.6, 4.0, 2.8, composition not
Leeds calculation
to last PMMA, 108 2.0, 1.4, 1.0, specified, con-
TO10 method
disc diameter 0.70, 0.50, trast range 0.011
not specified
0.35, 0.25 to 0.954 @ 70
kV, 1 mm Cu

17 thicknesses,
thickness and
Discs, 11.1, 7.9,
not specified, AutoPIA
Count 10 mm 5.6, 4.0, 2.8,
Leeds composition calculation
to last PMMA, 108 2.0, 1.4, 1.0,
TO12 not specified, method
disc diameter 0.70, 0.50,
contrast range not specified
0.35, 0.25
0.0043 to 0.954
@ 70 kV, 1 mm Cu
20 thickness,
thickness not
Discs, 11.1, 7.9,
specified, AutoPIA
Count 10 mm 5.6, 4.0, 2.8,
Leeds composition calculation
to last PMMA, 144 2.0, 1.4, 1.0,
TO20 not specified, method
disc diameter 0.70, 0.50,
contrast range not specified
0.35, 0.25
0.0016 to 0.954
@ 70 kV, 1 mm Cu

DIN 6868- Various e.g.

On-line data
15 [Link] 310 x 310
Count 8 thicknesses logging
NORM RAD/ mm, thick- Discs,
to last 8 (contrasts 0.9% available
FLU or ness 30 mm 10 mm
disc to 9.4%) e.g. at Pro-
Pro-Project PMMA with
[Link]
Pro-Fluo 150 1.5 mm Cu

version 01.2024 67
NEMA XR21 – Count Disc/cylinder Iodine embedded
25 mm
2000 cardiac to last 32 targets, 4, 3, in epoxy, 20, 10, 5 No
PMMA
phantom disc 2, 1 mm and 2.5 mg/cm²

Line pair test objects

A test object measurement of limiting spatial resolution is made using a line pair test object, which consists of
lines separated by certain spacing, corresponding to a spatial frequency, cut into a thin lead test object. The
lines are grouped by spatial frequency and the last spatial frequency group that can be resolved is defined
as the limiting or threshold resolution. In fact, the test object generates ~square waves, which include a
fundamental sinusoidal wave for each spatial frequency group.

The line pair test object gives an estimate of the highest spatial frequency that is transmitted through the
imaging system (provided this fundamental frequency exists in the test object). As with the TCDD, the factors
influencing the limiting spatial resolution depend on the position of the test object when imaged. When
positioned at the detector input plane, the x-ray detector is the determining factor. For detector tests, a high
dose rate (without saturation) and low energy to limit the influence of noise and contrast (signal) on the
result. Flat panel detectors have a fixed pixel spacing, with binning or interpolation applied in some FOVs, and
the limiting resolution will generally follow the Nyquist frequency i.e., Ny = 1/(2.p) where p is the pixel spacing.
This is the case provided that visual bias is not limiting the measurement i.e. the test object is sufficiently
magnified at the display stage when being scored. Limiting spatial resolution of the x-ray detector is a useful
routine test for XRII-TV based fluoroscopy systems where XRII focus can change due to faults or drift in the
focusing voltages.

When the test object is positioned at the patient position, geometric blurring from the x-ray tube will also
influence the result and in fact noise and x-ray contrast could also have some influence on the result. Used
like this, the line pair test becomes a system sharpness test.

Again, visual reading is often used when scoring this test object, even though digitally stored images are gen-
erally available now via a fluoroscopy store function. Some caution is required when visually scoring a line pair
test object acquired on digital image receptors as aliasing can generate additional patterns within a given line
pair frequency (Albert et al., 2002). Should digital images be available without re-binning or downsampling,
then the square wave contrast transfer function can be calculated, which gives a quantitative estimate of
sharpness and can be used as an alternative to an MTF measurement (Droege and Morin, 1982). It should be
noted that measurement of the MTF gives a more complete and robust evaluation of detector sharpness.

Line pair test object phantoms

The most commonly used line pair test objects are manufactured by Hüttner Röntgenteste (DE) (Table Ap-
pendix 2-2), where line pairs are cut into a thin Pb sheet. Different Pb sheet thicknesses are available, from
0.01 mm Pb to 0.1 mm Pb, with 0.1 mm Pb being a common thickness. There is a 12% change in spatial frequen-
cy between each spatial frequency group in these test objects. The Type 18 and Type 38 phantoms are often
included in test objects designed by other test object manufacturers. Hüttner Type 18 is used in the Leeds
TOR18 FG. Hüttner Type 38 (or a version of this test object) is used in the Leeds Fluoro 4, the PTW/Freiburg
NORMI 13, the RaySafe Pro-Fluoro 150, the Pro-Fluo 150 from Pro-Project and the NEMA XR-21 test objects.

68 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Table Annex 2-2. Hüttner line pair test objects

Base plate
# line Line pair Scoring
Name Composition / Image/diagram
pair groups frequencies software
thickness

Hüttner Pb/ 0.01 mm 0.5 lp/mm

21 No
type 18 to 0.1 mm to 5.0 lp/mm

Hüttner Pb/ 0.01 mm 0.6 lp/mm

20 No
type 18 to 0.1 mm to 5.0 lp/mm

Digital subtraction angiography (DSA)

Digital subtraction angiography (DSA) is an imaging mode designed to improve the visualisation of contrast
within blood vessels by removing overlying structures. The technique involves the injection of an iodine-based
contrast agent in the vessel of interest. Image acquisition starts just prior to the injection and continues until
sufficient contrast has been delivered to visualise the vessel or arterial/venous structure. A mask image is
defined using one of the images acquired before the contrast is introduced and this image is subtracted from
subsequent images in the sequence. If the patient does not move then overlying structures common to both
images cancel out leaving only the difference between the mask and the later images i.e. the filled vessel
containing the iodine contrast. A logarithmic transform is applied to the images before the subtraction is
performed, as this stops the modulation of the filled vessel by the signal intensities in the overlying anatomy.
Subtracting two images that contain uncorrelated (quantum) noise increases the variance of the noise in the
resulting (subtracted) image by a factor of 2 and therefore DSA modes often have increased exposure/image
to reduce noise. Other options to control noise in the subtracted images include image processing and the
use of a higher exposure/image for the mask images.

Evaluation of these imaging modes is more complicated than for non-subtracted imaging. A phantom con-
taining a blank mask is positioned in the FOV, the DSA is started and the blank mask removed and replaced by
a detail plate containing objects or details of interest. Clearly, manipulation of the test objects in the primary
beam and close to the presence of a scattering object must be done carefully.

A number of TIQ test objects are available for testing DSA modes. Leeds test objects supply the TO20 TCDD
(Cowen, Haywood, et al., 1987; Cowen, Workman, et al., 1987), which has a similar layout and covers the same
diameter ranges as TO10 and TO12 but can be used to test smaller contrasts, with 144 discs in total. The discs
are not made from iodine or iodine simulating material. The Leeds test objects have linear and logarithmic
step wedges for testing the accuracy and reproducibility of DSA system look up tables. The now defunct IEC
standard 61223-3-3 described a DSA performance testing phantoms which are still produced e.g. X-Check
DSA test object (PTW), QUART DSA phantom (QUART), Leeds DSA 8/54 test object (see Figure Appendix
2-3). The test object has a 57 mm thick PMMA body with a 10 mm slot for a 9.5 mm thick PMMA slider. This
slider contains aluminium strips 10 mm in diameter with thicknesses of 0.05, 0.1, 0.2 and 0.4 mm, simulating

version 01.2024 69
iodine contrast concentration of 5 to 10 mg/ml for vascular diameters of 1 to 4 mm. There is also a copper step
wedge, with 7 steps from 0.2 to 1.4 mm Cu, to test dynamic range and the logarithmic look up table (LUT)
applied in DSA mode. These test objects feature a pneumatic hand pump for safe movement of the mask and
detail plates during the DSA run.

Figure Appendix 2-2. Examples of DSA test objects designed to meet the previous IEC Standard IEC 61223-3-3 (left) Quart
(DE) DSA test object (right) PTW –Check DSA test object.

Tests of DSA performance are currently not included in this QC protocol but may be added in the future.
One could question the value of an explicit test of DSA modes, and whether these modes could be tested by
applying a standard TIQ method to acquire a DSA sequence and then evaluating the images without applying
the subtraction. A potential problem with this is that DSA modes tend to use a higher exposure/image, which
means that system structured (fixed pattern) noise is multiplied more strongly into the image and this can
limit the measured threshold contrast (Kume et al., 1986; Marshall et al., 2001). This could potentially bias
the result and underestimate the TIQ score for the DSA modes. An explicit evaluation of a commonly used
DSA mode especially given that aspects such as noise reduction and increased exposure/image for the
mask could influence TIQ and eventually clinical image quality. Physicists can apply one of the four TCDD
methods described (Section 4.1) however some of these methods have not been validated for DSA imaging.
Furthermore, DSA modes typically operate at between 2 and 4 images/second, meaning that Methods 1 and
3, requiring the acquisition of 400 or 1000 images respectively, are not practical options.

70 QUALITY CONTROL OF DYNAMIC X-RAY IMAGING SYSTEMS / EFOMP PROTOCOL

Temporal/motion test objects
Given that fluoroscopy and angiography devices are designed to image moving structures it is surprising
that so few test objects have been designed and implemented to evaluate the dynamic performance of
these systems. An exception is the work by Guibelalde et al ((Guibelalde et al., 2001, 2004; Kotre, Marshall
and Guibelalde, 2005) where a programmable moving platform was used to study the influence of motion
on test object scores in dynamic imaging systems. Difficulty in scoring a moving test object visually may
account for the sparse work on this topic, as digital images were unavailable on early fluoroscopy units. A
second problem is designing a test object with relevant motion and targets and relating the results to clinical
performance, a problem with TIQ test objects in general.

A rotating spoke test object has been developed (Model L-629, Ludlum Medical Physics, USA), which is
described for use in the NEMA XR-21 cardiac phantom (Balter et al., 2001). A circular acrylic disk of diameter
~14 cm contains 12 steel wires arranged in a spoke pattern at 30° intervals, along with 6 lead dots at the end
of each wire. The wire diameters range from 0.13 mm to 0.56 mm. The disk is rotated at 30 revolutions per
minute (RPM) in order to simulate the motion of guidewires etc. Details of how to evaluate the phantom are
not given and there are no published papers reporting application of this test object or giving typical values.

Figure Appendix 2-3. The Ludlum rotating spoke test object (left) An x-ray image of the test object showing the presence of
lag (the lower, faint line next to Line 1 in this image) (Balter et al., 2001).

Another option is to use a platform to move a test object at some velocity, a method that has been applied
to the measurement of the MTF (Friedman and Cunningham, 2008; Dehairs et al., 2017; Monnin et al., 2021).
Leeds Test Objects supply a moving platform (MOTUS) that can be programmed for speeds between 0 and
7.5 cm/s with a distance range of 5 cm.

This report does not describe an explicit assessment of the temporal imaging performance of fluoroscopy
using test objects, however it is recognized that this is a very important aspect of system performance,
especially for cardiac imaging systems. Updates of this work may include such an evaluation.

version 01.2024 71
QUALITY
CONTROL OF
DYNAMIC X-RAY
IMAGING
SYSTEMS
EFOMP PROTOCOL
VERSION 01. 2024

[Link]