Artificial Intelligence-Based

Infrared Thermal
Image Processing and
Its Applications
Infrared thermography is a fast and non-invasive technology that provides a
map of the temperature distribution on the body’s surface. This book provides
a description of designing and developing a computer-assisted diagnosis (CAD)
system based on thermography for diagnosing such common ailments as rheuma-
toid arthritis (RA), diabetes complications, and fever. It also introduces applica-
tions of machine-learning and deep-learning methods in the development of CAD
systems.
Key Features:

• Covers applications of various image processing techniques in thermal

imaging applications for the diagnosis of different medical conditions
• Describes the development of a computer diagnostics system (CAD) based
on thermographic data
• Discusses deep-learning models for accurate diagnosis of various diseases
• Includes new aspects in rheumatoid arthritis and diabetes research using
advanced analytical tools
• Reviews application of feature fusion algorithms and feature reduction
algorithms for accurate classification of images

This book is aimed at researchers and graduate students in biomedical engineer-

ing, medicine, image processing, and CAD.
Artificial Intelligence-
Based Infrared Thermal
Image Processing and
Its Applications

U. Snekhalatha
K. Palani Thanaraj
Kurt Ammer
First edition published 2023
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742

and by CRC Press

4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
CRC Press is an imprint of Taylor & Francis Group, LLC

© 2023 U. Snekhalatha, K. Palani Thanaraj and Kurt Ammer

Reasonable efforts have been made to publish reliable data and information, but the author and
publisher cannot assume responsibility for the validity of all materials or the consequences of their use.
The authors and publishers have attempted to trace the copyright holders of all material reproduced in
this publication and apologize to copyright holders if permission to publish in this form has not been
obtained. If any copyright material has not been acknowledged please write and let us know so we
may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, access [Link]
or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. For works that are not available on CCC please contact mpkbookspermissions@tandf.
[Link]

Trademark notice: Product or corporate names may be trademarks or registered trademarks and are
used only for identification and explanation without intent to infringe.

ISBN: 978-1-032-15814-3 (hbk)

ISBN: 978-1-032-15817-4 (pbk)
ISBN: 978-1-003-24578-0 (ebk)
DOI: 10.1201/9781003245780

Typeset in Times
by SPi Technologies India Pvt Ltd (Straive)
Contents
Preface������������������������������������������������������������������������������������������������������������������ xi
Acknowledgments����������������������������������������������������������������������������������������������xiii
Authors���������������������������������������������������������������������������������������������������������������� xv

Chapter 1 Fundamentals of Infrared Thermal Imaging.................................... 1

1.1 Basics of Thermometry......................................................... 2
1.2 Thermal Properties of Biological Tissues.............................. 3
1.3 Thermal Signals Associated with Physiological
Functions............................................................................... 4
1.3.1 Temperature Signal of Respiration........................... 5
1.3.2 Thermal Heart Rate Signal....................................... 9
1.3.3 Identification of Sweat........................................... 11
1.3.4 Facial Muscle Activation Signals........................... 14
1.4 Examples of Recent Applications of Infrared Thermal
Imaging in Medicine............................................................ 16
1.4.1 Rheumatoid Arthritis.............................................. 16
1.4.2 Diabetes Mellitus.................................................... 18
1.4.3 Dermatology........................................................... 18
1.4.4 Dental Disorders..................................................... 19
1.4.5 Complex Regional Pain Syndrome........................ 21
1.4.6 Fever Screening...................................................... 23
1.5 Summary.............................................................................. 25
References...................................................................................... 26

Chapter 2 Protocol for Standardized Data Collection in Humans.................. 29

2.1 Background and International Guidelines........................... 29
2.2 Thermal Camera Performance Check................................. 30
2.3 Ambient Temperature Control............................................. 33
2.3.1 Choice of Ambient Temperature............................ 38
2.3.2 Structure of the Examination Space....................... 38
2.4 Subject Selection................................................................. 40
2.4.1 Procedures Performed during the Examination..... 41
2.4.2 Behavior Prior to the Thermal Imaging.................. 42
2.4.3 Data Checking at the Beginning of the Imaging
Session.................................................................... 42
2.5 Patient Position and Image Acquisition............................... 43
2.5.1 Static or Dynamic Thermal Imaging...................... 45
2.6 Thermal Image Analysis...................................................... 47
References........................................................................... 50

v
viContents

Chapter 3 Basic Approaches of Artificial Intelligence and Machine

Learning in Thermal Image Processing.......................................... 55
3.1 Image Source....................................................................... 55
3.2 Data Transfer from Camera to the Analyzing Software...... 55
3.3 Pre-processing..................................................................... 56
3.3.1 Thermal Image Color Modes................................. 57
3.3.2 Image Enhancement............................................... 58
[Link] Image Smoothening and Sharpening...... 58
[Link] Histogram Equalization.......................... 59
3.3.3 Multispectral Images.............................................. 60
3.3.4 Image Registration................................................. 60
3.3.5 Image Fusion.......................................................... 62
3.4 Segmentation Techniques.................................................... 63
3.4.1 Clustering-Based Segmentation Algorithms.......... 64
[Link] k-Means Clustering................................. 64
[Link] Hierarchical Clustering........................... 65
[Link] Divisive Clustering................................. 65
[Link] Density-Based Clustering....................... 66
[Link] Fuzzy C-Means Clustering..................... 66
[Link] Neutrosophic C-Means Clustering......... 67
[Link] Mean Shift Clustering............................. 67
3.4.2 Threshold-Based Segmentation.............................. 67
3.4.3 Region-Based Segmentation Algorithms............... 68
[Link] Region Growing...................................... 69
[Link] Region Merging...................................... 69
3.4.4 Active Contour....................................................... 69
3.4.5 Edge Detection for Image Segmentation................ 70
[Link] Robert Edge Detection............................ 70
[Link] Sobel Edge Detection............................. 70
[Link] Prewitt Edge Detection........................... 71
[Link] Marr–Hildreth Edge Detection............... 72
[Link] Canny Edge Detection............................ 72
3.4.6 Watershed Segmentation........................................ 73
3.4.7 Deep Learning Models for Image Segmentation.... 73
3.5 Artifact Suppression............................................................ 73
3.6 Feature Extraction and Classification.................................. 74
3.6.1 Gray Level Co-occurrence Matrix.......................... 74
3.6.2 Speeded-Up Robust Features................................. 75
3.6.3 Gray Level Run Length Matrix.............................. 76
3.6.4 Local Binary Pattern............................................... 77
3.6.5 Gray Level Size Zone Matrix................................. 77
3.6.6 Local Binary Gray Level Co-occurrence Matrix...... 78
3.6.7 Scale-Invariant Feature Transform......................... 79
3.6.8 Local Directional Pattern........................................ 79
Contents vii

3.6.9 Segmentation-Based Fractal Texture

Analysis.................................................................. 80
3.7 Machine Learning Classifiers.............................................. 80
3.7.1 Logistic Regression................................................ 80
3.7.2 Naïve Bayes Classifier............................................ 81
3.7.3 K-Nearest Neighbors.............................................. 81
3.7.4 Decision Tree.......................................................... 81
3.7.5 Support Vector Machine......................................... 82
3.7.6 Random Forest....................................................... 82
3.7.7 Bagging.................................................................. 83
3.7.8 Logit Boost............................................................. 83
3.7.9 K-Star..................................................................... 83
3.8 Deep Learning Classifier..................................................... 84
3.8.1 Background............................................................ 84
3.8.2 Basic Architecture.................................................. 85
[Link] Convolution layer.................................... 85
[Link] Rectified Linear Unit (ReLU)................. 86
[Link] Pooling Layer.......................................... 86
[Link] Max Pooling............................................ 86
[Link] Average Pooling...................................... 86
3.8.3 Batch Normalization.............................................. 87
3.8.4 Drop Out................................................................. 87
3.8.5 Fully Connected Layer........................................... 87
3.8.6 Training Procedure for CNN.................................. 87
3.8.7 Pre-Processing and Augmentation of Data............. 87
3.8.8 Initialization of Parameters.................................... 88
[Link] Random Initialization............................. 88
[Link] Unsupervised Pre-Training
Initialization............................................ 88
3.8.9 CNN Regularization............................................... 88
3.8.10 Choosing an Optimizer........................................... 89
[Link] Gradient Descent.................................... 89
[Link] AdaDelta................................................. 89
[Link] Adam....................................................... 89
3.9 Performance Metrics........................................................... 89
3.9.1 Confusion Matrix................................................... 89
3.9.2 Accuracy................................................................. 90
3.10 Pre-Trained Network Models.............................................. 91
3.10.1 VGG Architecture................................................... 91
3.10.2 ResNet V2.............................................................. 92
3.10.3 DenseNet 121 (Densely Connected
Convolutional Networks)....................................... 92
3.10.4 MobileNet............................................................... 93
3.11 Conclusion........................................................................... 93
References...................................................................................... 94
viiiContents

Chapter 4 Thermal Imaging for Arthritis Evaluation in a Small

Animal Model................................................................................ 97
4.1 Induction and Evaluation of Arthritis.................................. 97
4.1.1 Collagen-Induced Arthritis..................................... 97
4.1.2 Mono-Arthritis Model.......................................... 100
4.1.3 Pristane-Induced Arthritis.................................... 104
4.1.4 Adjuvant-Induced Arthritis................................... 105
4.1.5 Antigen-Induced Arthritis.................................... 106
4.1.6 Proteoglycan (PG)-Induced Arthritis................... 106
4.1.7 Carrageenin Induced Arthritis.............................. 106
4.2 Thermal Image Acquisition............................................... 107
4.3 Thermal Image Analysis.................................................... 108
4.4 Histopathology Evaluation................................................ 112
4.4.1 Histopathological Analysis of Synovial Fluid...... 112
4.5 Conclusion......................................................................... 118
References.................................................................................... 119

Chapter 5 Thermal Imaging for Inflammatory Arthritis Evaluation............. 121

5.1 Patients and Methods......................................................... 122
5.2 Disease Activity Assessment............................................. 122
5.3 Temperature Indices.......................................................... 124
5.3.1 Heat Distribution Index........................................ 125
5.4 Thermal Image Acquisition Protocol................................. 126
5.5 Thermal Image Segmentation............................................ 130
5.5.1 Threshold Method................................................ 131
5.5.2 Edge Detection..................................................... 132
5.5.3 k-Means Clustering.............................................. 135
5.5.4 Watershed Segmentation...................................... 136
5.6 Statistical Feature Extraction and Classification............... 138
5.6.1 Oriented Fast and Rotated BRIEF........................ 138
5.6.2 Support Vector Machine....................................... 139
5.6.3 k-Nearest Neighbor.............................................. 140
5.6.4 Linear Discriminant Analysis (LDA)................... 140
5.6.5 Random Forest Classifiers.................................... 141
5.6.6 Naïve Bayes Classifier.......................................... 142
5.7 Findings and Conclusion................................................... 143
5.8 Case Study......................................................................... 144
5.8.1 Color Doppler Ultrasonographic Evaluation........ 146
5.8.2 Case Reports......................................................... 147
[Link] Case Study 1......................................... 147
[Link] Case Study 2......................................... 148
[Link] Case Study 3......................................... 149
5.8.3 Results of CDUS Examination............................. 149
Contents ix

5.8.4 Comparison of Clinical, Thermographic,

and CDUS Findings............................................. 150
5.8.5 Discussion............................................................ 150
5.8.6 Limitations and Recommendations...................... 151
References.................................................................................... 152

Chapter 6 Potential of Thermal Imaging to Detect Complications in

Diabetes: Rationale for Diabetes Screening with
Thermal Imaging.......................................................................... 155
6.1 Introduction....................................................................... 155
6.1.1 Diagnosis.............................................................. 156
6.1.2 Thermography in Diabetes Evaluation................. 158
6.2 Selection of Study Population........................................... 159
6.2.1 Subject Preparation.............................................. 160
6.3 Measurement Protocol....................................................... 161
6.4 Study of Thermograms in Subjects with Diabetes............ 161
6.4.1 Facial Thermal Imaging Measurements............... 162
[Link] Temperature Measurement
Analysis................................................ 162
6.4.2 Thermal Imaging in Diabetic Foot....................... 166
6.5 Computer-Aided System for Screening of Diabetic
Complications.................................................................... 173
6.5.1 Features................................................................ 174
[Link] Feature Extraction and
Classification........................................ 174
[Link] Oriented Fast and Rotated Brief
(ORB)................................................... 174
6.5.2 Automated Deep Learning Classifier................... 176
[Link] Pre-trained CNN Classifier................... 176
6.6 Conclusion......................................................................... 177
References.................................................................................... 178

Chapter 7 Thermal Imaging in Detection of Fever for Infectious

Diseases........................................................................................ 181
7.1 Human Body Temperature................................................. 181
7.1.1 Fever Development Due to Infectious
Diseases................................................................ 184
[Link] Thermal Imagers for Elevated Body
Temperature Detection.......................... 184
7.1.2 Measurement Protocol.......................................... 186
7.2 Fever Detection Systems................................................... 191
7.2.1 Conventional Methods.......................................... 191
xContents

7.2.2 AI-Based Fever Screening Systems..................... 192

[Link] Face and Eye Detection........................ 193
7.3 Conclusion......................................................................... 200
Acknowledgment.......................................................................... 201
References.................................................................................... 202

Chapter 8 Ethical Aspects in Thermal Imaging Research............................. 205

8.1 Introduction....................................................................... 205
8.2 General Ethical Principles................................................. 207
8.3 Guidelines for Good Clinical Research Practice............... 208
8.4 Responsible Conduct of Research..................................... 211
8.5 Ethical Issues Review Procedure....................................... 211
8.6 Informed Consent Process................................................. 212
8.7 Issues Related to Medical Datasets................................... 212
8.8 General Guidelines When Using Thermal Imaging for
Clinical Screening............................................................. 213
8.9 Ethical Considerations in Thermal Imaging Research...... 214
8.9.1 Infrared Imaging Systems as Medical
Devices................................................................. 214
[Link] Standards of Thermal imaging
system................................................... 215
[Link] Handling Thermal Imaging Systems.... 217
8.9.2 Patient Management Protocol.............................. 218
[Link] The Pre-Examination Process............... 218
8.9.3 Interpretation and Reporting................................ 219
8.10 Summary............................................................................ 220
Appendix I.................................................................................... 221
Appendix II.................................................................................. 223
References.................................................................................... 224
Index����������������������������������������������������������������������������������������������������������������� 225
Preface
Body temperature is a vital parameter used in the medical field for indicating
abnormal activity of human tissues. Infrared (IR) thermal imaging produces
information on skin temperature distribution in various body regions. The thermal
property of human skin varies between the different layers or within the layers of
the skin due to the presence of the blood vessels. Depending on the individual’s
condition and the surrounding environment, the human body will dissipate the
excess heat energy. Over the decades, many research works have been carried out
to analyze this temperature profile on various skin regions to identify the physi-
ological process behind it. This procedure forms the basis of thermal imaging,
whereby advanced thermal imaging sensors are used to capture the heat profile
of the human body regions so as to identify the ailments in the human body.
With the advancement in thermal imaging systems, high-resolution thermograms
could be produced with a temperature sensitivity of less than 50 mK and with a
spatial accuracy of around 10 μm. Widespread research is going into developing
computer-aided disease diagnosis systems based on IR thermograms as an assist-
ing tool for physicians in clinical screening and evaluation process. The main aim
of this book is to provide a detailed description of designing and developing a
computer-assisted diagnosis (CAD) system based on thermography for diagnos-
ing some of the common ailments such as rheumatoid aArthritis (RA), diabetes
complications, and fever. This book also introduces the readers to applications
of machine-learning and deep-learning methods in the development of the CAD
system.
This book is organized as follows:

• Chapter 1 deals with the fundamentals of infrared thermal imaging, which

includes thermal imaging properties, image acquisition, and various
applications.
• Chapter 2 focuses on the standardized protocol to be followed in the selec-
tion of test subjects and data acquisition based on infrared thermographs.
• Chapter 3 provides information on image analysis with a focus on image
analysis tools and various image processing steps involved in thermal
imaging.
• Chapter 4 evaluates arthritis disease progression in the CFA-induced Wistar
rat model and monitors inflammatory arthritis activity using thermal imag-
ing compared with histopathological study.
• Chapter 5 showcases the applications of thermal imaging in the diagnosis of
rheumatoid arthritis in humans.
• Chapter 6 explains the recent developments in applications of thermal
imaging in screening complications in diabetic subjects and discusses their
rationales.

xi
xiiPreface

• Chapter 7 explains the development of an advanced thermal imaging system

that uses deep learning approaches to detect elevated body temperatures for
proper fever screening.
• Chapter 8 describes the ethical considerations and data protection in ther-
mal imaging research.
Acknowledgments
First and foremost, I express our heartfelt and deep sense of gratitude to
Chancellor Shri. T.R. Pachamuthu, Vice-Chancellor Dr. C. Muthamizhchelvan,
Pro-Vice-Chancellor (Admin) Dr. Ravi Pachamoothoo, and Pro-Vice-Chancellor
(Academics) Dr. P. Sathyanarayanan of SRM Institute of Science and Technology
for providing us the necessary facilities for the completion of our book. I also
acknowledge Registrar Dr. S. Ponnusamy of SRM Institute of Science and
Technology for his constant support and endorsement.
I wish to express our sincere gratitude to our Dean (Engineering & Technology,
Kattankulathur) Dr. T.V. Gopal, and Chairperson (School of Bioengineering) Dr.
M. Vairamani of SRM Institute of Science and Technology for their constant sup-
port and encouragement.
I am extremely grateful to the Head of the Department, Dr. Varshini Karthik,
Biomedical Engineering Department, SRM Institute of Science and Technology,
for her invaluable guidance, motivation, and timely and insightful technical dis-
cussions. We are immensely grateful for her constant encouragement, smooth
approach throughout our journey and to make this work possible. Also, I would
like to thank researcher scholars Ms. Richa Rashmi, Ms. Ahalya, Ms. Nithya
Kalyani, and my students Ms. Bhargavi and Ms. Sowmiya of Biomedical
Engineering Department, SRM Institute of Science and Technology, for their help
and suggestions.
I wish to extend my sincere thanks to Dr. J. Kumar, Department of General
Medicine, and Dr. Panchapakesa Rajendran, Head, Rheumatology Department,
SRM Hospital and Research Center, for providing support and encouragement in
collecting the data at SRM Medical College Hospital and Research Center.
I would to thank my family – my husband M. Ravishankar and his parents, my
parents, my brothers, and my children – for their continuous love, support, encour-
agement, and understanding. I thank the Almighty for his blessings.
U. Snekhalatha
SRM Institute of Science and Technology

I would want to take this opportunity to show my gratitude to everyone who has
helped us finish this book; to our All-Powerful God for providing us with the
strength, knowledge, patience, and wisdom necessary to complete this book.
I would like to express my strong heartfelt gratitude to the Management of St.
Joseph’s College of Engineering, Chennai, India, for their constant support and
encouragement,
I would also want to express our heartfelt gratitude to our colleagues, friends,
research scholars, and instructors who have been kind enough to provide recom-
mendations and comments for the text's improvement. Additionally, we would

xiii
xivAcknowledgments

like to express my gratitude to the individuals who provided support and presence,
as well as ideas for the development of this book.
K. Palani Thanaraj
St. Joseph’s College of Engineering

I like to thank Dr. U. Snekhalatha and Dr. K. Palani Thanaraj for inviting me to
contribute to this book. It was a pleasure to work with them, and their open mind
for improving this book is much appreciated. I am also grateful for their sup-
port and help in editing and correcting the text during a difficult personal health
condition.
Prof. Dr. med. Kurt Ammer, PhD
Editor in Chief, Thermology International
Treasurer, European Association of Thermology
Authors
Dr. U. Snekhalatha is currently working as an Professor in
the Department of Biomedical Engineering, SRM Institute
of Science and Technology (SRMIST), Kattankulathur,
India. She pursued her Doctorate in Biomedical
Engineering at SRMIST (2015). Her areas of interest
include biomedical signal processing, medical image pro-
cessing, biomedical instrumentation, and machine learn-
ing and deep learning techniques. She has published 88
research articles in reputed peer-reviewed international
journals and international conferences. She has filed four Indian patents of
which, all are in published stage. She obtained the Best Researcher Award for
her publications in Nature indexed journal during the Research Day function
held on March 1, 2021 at SRMIST. She received Gandhian Young Technological
Innovation Award for paper titled “Design of Acetone Breath Gas Analyzer in
the Evaluation of Diabetes Mellitus,” Appreciation 2020, Society for Research
and Initiatives for Sustainable Technologies and Institutions, November 5, 2020.
She has obtained best paper award and gold medals for some research paper
publications. She is a life member of various professional societies such as the
Biomedical Society of India, IEI, IEANG, IRED, ISTE, and ISCA. She is cur-
rently serving as a reviewer and guest editor for various reputed peer-reviewed
international journals.

Dr. K. Palani Thanaraj is currently working as an

Assistant Professor in the Department of Electronics
& Instrumentation Engineering, St. Joseph’s College
of Engineering, Chennai. He has completed his
PhD in 2018 in the Faculty of Information and
Communication Engineering from Anna University,
Chennai. His research areas include image process-
ing, advanced signal processing, image segmentation,
machine learning, and deep learning. He has devel-
oped deep learning algorithms for performing image
classification of medical images for disease diagnosis.
He has published his works in many reputed and ref-
ereed journals indexed in Web of Science and Scopus.

xv
xviAuthors

Prof Dr. med. Kurt Ammer was certified as a gen-

eral medical practitioner in 1978, a consultant for
physical medicine and rehabilitation in 1989, and a
consultant for physical medicine and rehabilitation
(rheumatology) in 1994. He was senior researcher at
the Ludwig Boltzmann Research Unit for Physical
Diagnostics, Austria, between 1988 and 2004. From
1985 until his retirement in early 2013, he was Vice
Director of the Institute of Physical Medicine and
Rehabilitation at the Hanusch Hospital in Vienna,
Austria. He got involved in medical thermography
in 1988, and was appointed as secretary and trea-
surer of the European Association of Thermology
(EAT) in 1990, and currently serves as the EAT treasurer. Since 2002, he has been
appointed as external professor at the Medical Imaging Research Unit, University
of South Wales, Pontypridd, UK. His research interests focus on rehabilitation
medicine and the application and standardization of thermal imaging in medicine.
 1 Fundamentals of
Infrared Thermal
Imaging

Infrared thermography is a non-invasive imaging modality which is used to deter-

mine the skin surface temperature of any object, since every object emits thermal
radiation when it is at a temperature above absolute zero. The heat content of
inanimate matter and living beings differs, especially if these living beings have
developed a physiological thermoregulation system leading to two different tem-
perature compartments – a central one with maintained temperature and a periph-
eral shell with constantly changing temperature reflecting the heat exchange with
the environment. It is essential to remember that temperature is one of the seven
base quantities of the International System of Units (SI). Like pressure and den-
sity, temperature is an intensive quantity. A physical property of a system that
does not depend on the system size or the amount of material in the system is
called intensive. Temperature can simply be seen as the “degree of hotness” or,
more specifically the “potential for heat transfer” (Machin and Tsai, 2010).
Like in other objects, heat transfer within and from the human body occurs
without a change of state involving conduction and convection. On the surface,
additional heat transfer is achieved by thermal radiation or by fluid evaporation;
the latter is combined with a change of state. Since the temperature is non-homo-
geneously distributed across tissues and blood vessels, there is constant heat
exchange between regions of different temperatures by conduction and convec-
tion. The tissue thermal resistance and bloodstream directions are the important
factors that determine the speed of heat transfer. Bloodstream and heat flow may
be in opposite directions, resulting in an additional “counterstream” mechanism
of heat exchange, uncontrolled by the thermoregulation system. The exchange of
heat from skin to air is caused mainly by means of infrared radiation. The radia-
tion emitted by the skin can be captured by a thermal imaging camera and dis-
played as radiance intensity equal to temperature. Since the anatomical details
obtained from the thermal imaging of the human body express the outline struc-
ture of the body, it is a common hypothesis that the physiological function in
health and disease are reflected by the temperature distribution on the skin.
Several research studies indicated that detection of diseases and disorders
could be possible by analyzing the thermal images of subcutaneous skin tempera-
ture. Thermal imaging has been used for detecting various medical conditions
affecting the locomotor and neuromuscular system such as rheumatoid arthritis
and other inflammatory arthritis, osteoarthritis, pain syndromes including

DOI: 10.1201/9781003245780-1 1
2 Artificial Intelligence-Based Infrared Thermal Image Processing

complex regional pain syndrome, orofacial pain, and diabetes complications.

Historically, the first medical application of thermal imaging was in breast cancer
detection. Among the early fields of thermographic research were also peripheral
vascular disorders such as Raynaud’s phenomenon or venous diseases, dermato-
logical disorders such as psoriasis or melanoma, and diseases of the locomotor
apparatus. Consecutively, the use of infrared thermography was extended to eye
diseases, psychological conditions, cardiovascular disorders, metabolic disorders,
and, more recently, to physical exercise assessment and mass fever screening.

1.1 BASICS OF THERMOMETRY
Thermometry may be defined as both the branch of physics dealing with the mea-
surement of temperature and the science of the construction and use of thermom-
eters. Physical perception of temperature is the basal experience related to heat
exposure. Phenomena such as dilation of air and vapor as a response to heat were
already known in ancient Greece, opening the opportunity to detect temperature
independent from the human temperature sense (Biró, 2011). Historically, detec-
tion and measurement of temperature are based on the laws of classical thermody-
namics. In particular, the zeroth law of thermodynamics describes the conditions
of thermal equilibrium, in which temperature is the parameter of a state that has
the same value for any systems which are in thermal equilibrium. This property is
used when the temperature of an object is measured by placing a thermometer in
contact with the object, initiating according to the second law of thermodynamics
heat flow from high to low temperature that finally results in thermal equilibrium
(Machin and Tsai, 2010). Biró (2011) described the ideal relationship between the
heat capacities of both the object to be measured and the thermometer to achieve
fast thermal equilibration. Other thermometers apply the thermoelectric effects
detected by Seebeck, Peltier, and Thompson as an indication of temperature values.
Thermometry can be classified based on the process of heat transfer involved
(conductive, convective, or radiative), the field of application (basic science,
applied science, industry, biomedicine, biology), or the temperature bands mea-
sured. Particularly, radiation thermometers are classified as high-, mid-, or low-
temperature measurement devices. Infrared thermal imagers are regarded as an
extension of low-temperature radiation thermometers (Zhang and Machin, 2010).
Each field of application has standards and limitations that vary between science
disciplines. For example, the ISO standard regulating the use and calibration of
clinical thermometers is different from the guideline controlling infrared cameras
for fever screening. Zhang and Machin described a schematic setup for radiomet-
ric temperature measurement. A source of infrared emission is the target for which
the temperature is to be measured. An optical system captures the input signal
infrared radiation for an infrared-sensitive detector which in turn produces an out-
put signal. The radiation-capturing system is constructed from a lens confining the
field of view, a bandpass optical filter determining the spectral region within
which the thermal radiation is to be measured, and an aperture in front of the
detector to avoid detection of stray light. The output signal (voltage or current)
Fundamentals of Infrared Thermal Imaging 3

from the detector is amplified and measured by a signal processor. This output
signal is related to the spectral radiance emitted by the surface and hence its tem-
perature can be deduced.
Clinical infrared radiometers were first developed in Germany in the 1920s
(Cobet, 1926). Most quantitative results of infrared-based clinical temperature
measurements in the early phase of infrared thermography were obtained using
spot thermometers. Using an optical–mechanical scanning system, thermal cam-
eras at that time were able to generate a monochrome image which was displayed
on a cathode ray screen. Since radiometric data were unavailable, infrared ther-
mography was in those days exclusively an imaging modality, and vague ideas
existed to analyze thermal images in a quantitative way.

1.2 THERMAL PROPERTIES OF BIOLOGICAL TISSUES

The thermophysical properties of both inanimate material and biological tissues
define the conditions for their conductive heat transfer. The properties include ther-
mal conductance defined as the quantity of heat that passes in unit time through a
plate of a particular area and thickness when its opposite faces differ in temperature
by one kelvin. For a plate of thermal conductivity k, area A, and thickness L, the
conductance is kA/L, measured in W·K−1. Thermal conductivity k, also symbolized
as λ or κ, is a material property, which may be derived from thermal conductance.
Thermal resistance Rk is the reciprocal of thermal conductance, L/kA, measured in
K·W−1. The ratio k/L is called conductive heat transfer coefficient, hk, measured in
W·K−1·m−2. The reciprocal of the heat transfer coefficient is thermal insulance, a
measurable quantity of importance for the calculation of heat loss.
“Heat capacity” c is a physical property of matter, defined as the amount of
heat to be supplied to an object to produce a unit change in its temperature. The SI
unit of heat capacity is joule per kelvin (J/K). Thus, heat capacity describes the
ability of matter to store thermal energy. Specific heat capacity cp is found by
dividing the heat capacity of an object by its mass. The subscript p in c points to
the condition of constant pressure, in which heat supplied to the system would
contribute to both the work done and the change in internal energy. Density ρ is
defined by the mass of a substance per unit volume. The product of density ρ and
specific heat capacity cp is called volumetric heat capacity ρ cp. Finally, thermal
diffusivity α combines thermal conductivity with density ρ and specific heat cp:
k
α= . At interfaces of contact, the product k ρ c represents an important fea-
ρcp
ture of the capacities for heat transfer in transient state. The square root of the
product of the material’s thermal conductivity and volumetric heat capacity is
called thermal inertia. The greater the thermal inertia, the higher the heat transfer.
Thermal inertia explains why the sensation of the temperature of a given object
depends not only on its actual temperature, but even more on its k ρ c values
(Houdas and Ring, 1982).
These physical properties can be used to calculate the passive heat transfer
from deep tissues to the surface of a living being. This is a process of physics and
4 Artificial Intelligence-Based Infrared Thermal Image Processing

not a control mechanism of physiology. It is important not to forget that the aim
of thermoregulation is to maintain deep body temperature within narrow regula-
tion. This is achieved by two strategies: changing the thermal resistance of the
skin and adaptation of metabolic heat production. Under constant blood pressure,
blood flow defined by the ratio of blood volume to tissue volume is dependent on
the width of the vascular beds, thus, vasoconstriction reduces blood flow and
increases thermal resistance and vasodilation acts vice versa. An additional mech-
anism for heat dissipation is the evaporation of sweat at the skin, a physiological
function available in some, but not all homeotherms. In humans, heat production
via cold-induced shivering is well developed, but reduced heat production such as
in hibernating mammals is unknown. The skin is an organ that acts as a boundary
and interface to the ambiance, but its temperature is not regulated. However, the
skin is the place where thermoregulatory responses take place and the site where
the heat transfer from deep body structures finally ends. Therefore, if any patho-
logic process is heat-producing, this thermal energy will be dissipated within the
body following the laws of physics, and a small amount of this heat may appear at
the skin depending on the location and the intensity of the heat source. Since ther-
moregulatory responses may interfere with temperature associated with disease
processes, it is of advantage to avoid any thermoregulatory responses during cap-
turing medical infrared images for diagnostic purposes.

1.3 THERMAL SIGNALS ASSOCIATED WITH PHYSIOLOGICAL

FUNCTIONS
At the skin, heat transfer to the environment is predominately through radiation,
which can be captured by an infrared camera and displayed as the temperature
on the site from where thermal energy is emitted. The resulting “static” thermal
image is analyzed by comparing temperature values obtained in the region(s) of
interest (ROI) with a reference temperature, which is often located in the corre-
sponding contralateral anatomical region. However, this temperature signal oscil-
lates in both the time domain and the spatial domain. Michael Anbar was one of
the first who subjected these temperature fluctuations using fast Fourier transform
(FFT) and relate the detected frequencies to other periodic phenomena such as
pulse or neuronal control of blood flow in the microvasculature (Anbar, 1990,
2013). Anbar also showed that the temporal change in surface temperature varies
in different anatomical regions.
Although the signal associated with physiological functions is the surface tem-
perature, its distribution is understood as an image feature, and not as a measur-
able quantity. Consequently, the models applied for extracting physiological
functions from surface temperatures are based on image processing.
Respiration contributes to the total heat exchange between the body and the
environment by the evaporation of humid air and by convective heat loss through
the breathing cycle. Insensible perspiration is the process of a small, but continuous
evaporative heat loss via the skin and the respiratory tract. At the skin, this loss is
due to a diffusion of water vapor, which can easily be detected by weighing the
Fundamentals of Infrared Thermal Imaging 5

body (Houdas and Ring, 1982). The predominant source of water content in the
skin is sweat production, mainly produced as a thermoregulatory response, but also
elicited by various pathologies leading to generalized, regional, or focal hyperhi-
drosis (Ammer and Ring, 2019). While the water content of the skin due to insen-
sible perspiration provides the basic condition of skin conductivity for direct
(galvanic) current, psychogenic sweating contributes to the variation of galvanic
skin response (GSR). The periodic blood pressure waves of the arterial pulse are
associated with skin temperature alterations. The temperature associated with per-
fusion is best seen at re-perfusion when external occlusion is released. Autonomic
control of microcirculation may be detected in thermal images of the face. This
approach attracted recently large interest in psychophysiology (Ioannou et al.,
2014), particularly, for the evaluation of emotions (Clay-Warner & Robinson, 2015).

1.3.1 Temperature Signal of Respiration

In general, inspired air is cooler and dryer than expired air. Air conditioning occurs
mainly in the upper airways and heat exchange is achieved in the small bronchi,
and air temperature equals the blood temperature of 37°C in the alveoli. At the nos-
trils, the temperature of expired air is slightly cooler than it was in the lungs. Thus,
the nostrils are a convenient site to obtain the temperature of both inspired and
expired air. In conventional respiratory measurements, a thermistor is placed near
the nostrils to measure the thermal breathing signal. In thermal imaging, the breath-
ing rate is estimated from the thermal signal detected in ROI located at the nostrils.
Most of the commercially available thermistors are not capable of measuring
the tidal volume and find difficulties in tracking the breathing rate accurately over
a wide range of breathing frequencies. But the study conducted by Lewis et al.
focused on an alternative method to evaluate and monitor the respiration rate and
relative tidal volume based on video recordings of facial thermal images. In their
study, an automated facial tracking model was used to measure the temperature
variations within and around the nostrils. Their obtained thermal signal is con-
verted to quantify the duration and relative depth of each breath cycle.
Lewis et al. (2011) developed a facial tracking algorithm to compensate for head
movements and estimated the dominant breathing rate by means of spectral analysis
of raw thermal signals obtained in infrared thermal videos of nostrils from 19 study
participants performing various breathing patterns. Dominant breathing rate was the
basis to calculate the breath-to-breath timing intervals and relative tidal volume.
Simultaneously to thermal imaging, respiratory inductance plethysmography was
used to estimate respiration rate and tidal volume. Statistically significant correla-
tions were found between the thermal signal and the plethysmograph measures such
as breathing interval and relative tidal volume.
The video tracking algorithms are used for automated extraction of thermal
signals and to compensate for heat movement. The tracking system has the advan-
tage of performing thermal video analysis, functioning in real time, and identify-
ing the nose features in 3D. Tao and Huang et al. developed the tracking algorithm
which is based on the piecewise Bezier volume deformation model (PBVD).
6 Artificial Intelligence-Based Infrared Thermal Image Processing

These PBVD algorithms could track the movement of particular facial features.
The PBVD model is comprised of a stack of 3D Bezier volumes embedded with
the face mesh model. The Bezier volumes are represented by two-layer, non-par-
allel polygons placed above and below the face mesh model. Image Pyramids
were constructed on each video frame and motion parameters were estimated. By
means of a tracking algorithm, the thermal signals were extracted from the ther-
mal video. The first frame from the thermal video is displayed and superimposi-
tion of 2D coordinates from all vertices in the 3D PBVD model is performed as
blue dots over the facial thermal image. Then, the pointer tool is used for the
selection of vertex points that enclose the nostrils defined as the ROI. Consequently,
an updated 3D model is generated for each video frame with a vertex containing
the selected points as a convex polygon.
The tidal volume signal was continuously recorded by using the life shirt (R)
inductive plethysmography. Thermal respiratory video signals were recorded for
three different breathing methods such as spontaneous, slow and deep, rapid, and
shallow. The respiratory signals were extracted from the thermal video signals
using the three-step filtering process. The dominant breathing rate was estimated
from the recording by means of spectral analysis. Initially, the extended length of
1024 samples was obtained by applying detrending and zero-padding operation.
Then, FFT is applied to measure the spectral density distribution. Due to the wide
range of breathing frequencies from 3 to 65 cycles/min, the frequency with the
highest spectral density was identified as the breathing rate.
As a second step, the thermal signal is converted into a time series signal which
is proportional to lung volume. This is achieved by defining the peak of the inspi-
ration signal as the cold temperature and that of the expiration signal as the warm
temperature. Thus, by integration of thermal time series, transformed time series
is formed which contains the components linearly related to tidal volume. The
integrated thermal time series signal was pre-processed using a dynamic filtering
process. To accomplish this process, a cubic-polynomial filter with a cut-off fre-
quency less than the breathing rate is considered. The filter cut-off frequency is
obtained by using the formula

fc = α ∗ BR (1.1)

where fc is the cut-off frequency, α = 0.385, and BR is the breathing rate of the
signal.
Fei and Pavlidis (2010) measured the breathing rate in thermal video using
automated tracking of the nasal region. They used coalitional tracking algorithm
for tracking the facial region during the breathing rate measurements. This can be
accomplished by using particle filter trackers which track the dynamic nature of
thermal imaging.
In a thermal image of a participant’s face, a coalition grid comprising 4 particle
filter trackers is placed. The grid outline is marked on the nose region of the facial
thermogram by using the click-and-drag method. It is considered as the ROI
initialization of the coalitional tracker.
Fundamentals of Infrared Thermal Imaging 7

The breathing rate is measured in the nostril region in thermal imaging. During
the inspiration process, the subject inhales the cool air from the outside environ-
ment. This leads to the convection of heat away from the nostril at an increased
rate. Thus, the nostril radiates the heat at a low level toward the camera. During
the expiration process, the subject exhales the warm air which is forcefully ejected
toward the environment. At this point, heat convection is weak due to the minimal
temperature difference observed between the expelled air and the nostril-expired
air. Thus, the nostril area radiates the heat at a high level toward the camera. The
authors segmented the nostril region from the facial thermograms. The horizontal
and vertical edges of the nostril region are detected by applying the Sobel opera-
tor. Then, spatial–temporal projections of the nostril in 3D space are obtained for
horizontal and vertical projections. Further, they computed the mean horizontal
and vertical integral projections for all the frames using the time window for both
expiration and inspiration phases. They determined the average temperature
within the measurement ROI in every frame. The wavelet transform is applied to
the thermal signal to obtain the breathing rate of the signal. Initially, the thermal
signal is sampled at a rate of δ = 10 fps. The normalization is applied over the
resampled thermal signals to normalize the signal amplitude. Then, continuous
wavelet transformation is performed on the normalized thermal signal by using
the sliding window. The breathing signal is extracted from the continuous wavelet
transform (CWT) analysis. Based on the wavelet scales, the breathing waveform
is computed from the nostrils. The breathing rate is estimated by using the center
frequency Fc of the mother wavelet and sampling factor δ = 10 fps.

Breathing rate = Fc δ (1.2)

Pavlidis measured the breathing rate of the respiratory signal from the expired
air. Another drawback of their method is that raw thermal data were used to build
up the signal for each pixel of desired ROI, which affects the signal fidelity due to
extreme noise. Hence, to overcome these disadvantages, Chekmenev et al. (2005)
indicates the thermal image at different scales and select the important scale which
carries relevant information with respect to breathing and heart signal. The authors
manually selected the carotid area of the neck as ROI-1 and the nasal area as ROI-
2. Then, they decomposed each frame of the thermal images into three average
value scales. They calculated the average value of ROI-1 and ROI-2 for each scale
of all the frames and obtained 1D plots for the average temperature value.
Further, they applied continuous wavelet analysis on 1D plots.
The CWT is defined as
∞
 t −b 
w(a,b) = 1/ √Cφ 1/√a ϕ ∗ 
−∞
∫
 a 
 f ( t ) dt

ϕ ^ (ω ) 2
∞

Cϕ =
∫
−∞
ω
dω < ∞ (1.3)
8 Artificial Intelligence-Based Infrared Thermal Image Processing

∫
ϕ ^ (ω ) = ϕ ( t ) e − jωt dt
−∞

φ(t) is mother wavelet.

 t −b 
ϕ a ,t ( t ) = ϕ  
 a 

where t indicates time, which can also be frame, and b is the time shift.
The reconstructed signal is obtained by using inverse continuous transform.
The average value of the signal is added to the inverse wavelet transform to
obtain the reconstructed signal. Based on the prior knowledge of the important
structure that belongs to the appropriate scale, the inverse continuous wavelet
transform (ICWT) analysis can be performed accurately. Finally, the reconstructed
signal can be obtained by adding the average value of the signal to the ICWT. The
heart rate is computed by counting the maxima in the reconstructed signal.
Jakkaw and Onoye (2020) monitored the respiratory activity and body move-
ments during the sleep using the thermal imaging technique. The breathing pat-
terns are detected during sleeping and identifying the sleep disorders such as sleep
apnea caused due to hypertension, cardiovascular diseases, and arrhythmia.
Hence, the authors demonstrated the non-contact method of respiratory and body
movement detection using a thermal camera which detects the temperature
changes due to breathing patterns. The thermal video frames of subjects in sleep-
ing mode are considered the input images. The images are pre-processed using the
Gaussian filter. In the respiration-monitoring method, ROIs are automatically
detected from the input images by means of identifying the highest temperature
point and massive portions of the high-temperature area. For automated ROI
detection of the thermal image, sleeping position is considered. The highest tem-
perature points are detected in the image by using minimum and maximum inten-
sities found in the image. The maximum pixel intensities are associated with the
highest temperature of the body. After determining the pixel at the center of the
observation area, a rectangular ROI of pixel size either 10 × 10, 25 × 25, or 50 ×
50 is applied. Among the ROI used, empirical research result shows that 50 × 50
pixels produced very good accuracy in compliance with an original frame 640 ×
480. The massive portions of the high-temperature area are detected using the
thresholding method. A threshold value of 176 was set, which produced best
results in segregating the human skin area from the background.
To extract the respiratory signals, the average of pixel values within each ROI of
every frame was calculated from the ROIh and the ROIt. The breathing motion is
detected by means of finding the difference between the current frame and the past
frame or previous frame. Then, the portion of the moved area is extracted using
thresholding and morphological operations such as erosion and dilation. Bounding
boxes are obtained by identifying the contours after filtering out the noise. The
breathing motion is counted based on the number of bounding boxes. The average
Fundamentals of Infrared Thermal Imaging 9

respiratory signal can be obtained as the combination of root-mean-square calcula-

tion of three signals attained from the ROIh, ROIt, and breathing motion.

Respiratory signal = √ ROI 2h + ROI 2t + BM 2 (1.4)

Then, these fused signals are passed to the Butterworth band pass filter with cut-
off frequencies of 0.05 and 1.5 Hz. The filtered signal is further smoothened by
using Savitzky–Golay (SG) filter. The moving average is applied to select only the
desired peaks. Then, the number of peaks counted depends on the number of breaths.

1.3.2 Thermal Heart Rate Signal

When the left ventricle pumps the blood from the heart to the other parts of the
body, a cardiac pulse is generated. The propagation of cardiac pulse involves
mechanical processes as the blood travels through the arterial network and turns
back to the heart via the vein network. Hence, Garbey et al. (2007) monitored the
pulse waveform by means of skin temperature modulation. This pulsative blood
flow modulated the skin tissue temperature due to heat exchange that occurs
between the vessels to surrounding tissue by means of convection and conduction.
Therefore, the authors proposed a model to simulate the heart diffusion process on
the skin’s superficial blood vessels and found that the skin temperature waveform
is analogous to the pulse waveform. They extracted the cardiac pulse from the
external carotid artery or the temporal arterio-venous complex or radial arterio-
venous complex. Thus, the modulation of temperature with respect to pulsating
blood flow generates the predominant thermal signal on a blood vessel. But this
signal is being affected by physiological and environmental factors. Hence, the
thermal signal captured by the thermal camera is a composite signal with the pulse
considered as one of its components.
In the pulse measurement method, as an initial step, a rectangular ROI is placed
over the blood vessel. Temperature profiles of the exposed blood vessel of the
subject are obtained. Then, the pixel temperatures along the horizontal direction
are averaged. This decreases the noise and constricts the rectangular ROI into a
line in which signal measurement initiated. Along the pixels on the measurement
line, time evolution signal of its temperature is obtained. Then, FFT is applied to
these temperature profiles to attain the power spectra. Then, power spectra of all
the profiles are averaged to yield the composite power spectrum.
Kim et al. (2018) proposed a non-invasive heart rate detection based on ther-
mal imaging in the blood vessels of the carotid artery. The blood vessel tempera-
ture varies with respect to the blood flow. Hence, it is necessary to identify the
blood vessel using a suitable algorithm. The authors used the binarization method
to detect the presence of blood vessels. After cropping the required ROI in the
thermal image, binarization is applied which produces the output in the form of
two-level intensities such as black and white. The background is represented in
black color and the foreground object is indicated in white color. Then, the mid-
point of the brightest region is measured to identify the position of blood vessels.
10 Artificial Intelligence-Based Infrared Thermal Image Processing

Then, they acquire the 1D time series signal at the vessel positions and remove the
noise in the signal using bit shift operation toward least significant bit (LSB)
detection. Hence, to detect the heart rate in real time, the time series signal is
converted into a frequency domain single using discrete Fourier transform (DFT).
The heart rate is calculated by choosing the frequency band of 0.9–1.6 Hz corre-
sponding to the heart rate of 54–96 beats/min. The final heart rate is obtained by
averaging the multiple heart rate from the multiple points on the identified blood
vessels. They compared the heart rate measured from PPG with that of changes in
temperature and found that blood vessel is identified correctly only when the esti-
mated accuracy is 95.48%. The accuracy of 78.65% was only achieved when the
blood vessels are not identified properly.
Pereira et al. (2018) estimated the heart rate from the thermal videos of
mechanical head movements co-occur with the cardiac cycle. As an initial step,
pre-processing operations are performed on the thermal images. A multi-level
thresholding algorithm was used to segment the head region from the background.
According to the algorithm, optimum threshold values are calculated based on the
discriminant analysis. The background elimination is followed by contrast
enhancement is carried out for all the frames of video thermograms. The next step
involves ROI selection particularly focused on the lower part of the head in each
frame of thermal video based on the semi-automated method. Then, feature points
are chosen from the desired ROI. The authors used Shi–Tomasi corner detector to
identify the optimal feature points within the ROI. According to the corner detec-
tor, the feature is considered as corner only when the scoring function R is
obtained based on the minimum eigenvalues λ1 and λ2 greater than the threshold.

R = min ( λ1, λ2 ) (1.5)

R >α

Initially, the feature points are identified in the first frame of the video sequence.
Then, the features are ranked according to the R-score and only the strongest fea-
tures were selected for tracking. The selected features were given as the input to
the template-based point tracker (Kannade Lucas Tomasi tracker) to position the
feature point exactly to the head movements. Hence, this tracker identifies all the
feature points from the first to the last frame of the video thermal image. As the
head movement occurs in the vertical direction, the vertical trajectories of each
single feature point over time were computed. After tracking the feature points,
temporal filtering is applied to detect vertical trajectories. In that, the low-fre-
quency components dominate the trajectories of feature points due to respiratory
movements, whereas high-frequency components provide accurate information
for detecting the peak of the heart rate signal. Apart from the vertical movement
of the head, other physiological sources such as vestibular activity, facial expres-
sion changes, and involuntary muscle movements contribute to the feature point
trajectories. Hence, to isolate the heart-related component of interest, it is
Fundamentals of Infrared Thermal Imaging 11

necessary to decompose other signals into sub-signals. Therefore, the researcher

introduced principle component analysis (PCA) to identify the important dimen-
sions with respect to the changes in the head position. They attained this by con-
verting the original dataset into a new data set of uncorrelated variables called
principle components. According to the PCA, an appropriate eigen vector is
selected to extract the pulse signal. The eigen vectors are ordered with respect to
their variance. The first principle component consists of the highest percentage of
data variation and ith principle component corresponds to ith highest amount of
variation. Hence, the first six principle components are considered for further
analysis in identifying the ballistocardiographic head movement. Then, FFT is
applied to obtain the amplitude spectra of principle components. As the principle
components are chosen based on the periodicity of the signal, the pulse signal is
extracted from the frequency components.

1.3.3 Identification of Sweat
Since evaporating sweat results in a local reduction of skin temperature, sweat
droplets appear in infrared thermal images as cold dots (Ring, 1984). In psycho-
genic sweating, sweat gland activation may persist in maxillary, nose tip, and
perioral regions.
Pavlidis et al. extracted the transpiration signal from thermal images in the peri-
nasal region (Pavlidis et al. 2012) and used the obtained signal as the primary indi-
cator of stress. They quantified the stress based on the estimation of transient
perspiratory responses using thermal imaging. The perspiration signal is obtained
using a virtual tissue tracker which tracks the ROI even with a few movements. The
tissue tracker has the tendency to operate on various head poses and thermal varia-
tions. A tracking algorithm is used for identifying the upper orbicularis oris portion
of the perinasal region in the initial frame. Spatial–temporal smoothing technique
was used in every frame of the thermal clip to estimate the best matching block. To
compute the perspiration signal, a morphology-based algorithm was used in each
ROI. FFT was used to suppress the high-frequency noise which occurs in the per-
spiration signal due to imperfections in tracking the signal and breathing effect.
As we know the perspiration signal produced from the perinasal region is the
primary indicator of stress, the activated perspiration pores appear as cold dark
spots in facial thermal imaging. The morphological operation used to display the
cold dark spot is black top-hat transformation. To avoid poor localization of per-
spiration spots, a contour-based structuring element is used. This contour-based
top hat transformation is used to capture the perspiration spots in every frame in
the thermal imaging. The instantaneous energy is calculated based on the time at
which the frame is captured and the number of detected cold spots at that time.
The evolution of instantaneous energy in the perinasal region produced an energy
signal which is indicative of perspiration activity.
Eccrine sweat glands are almost widely spread across the whole body. It
secretes watery fluid composed of water, NaCl and other solutes onto the skin
surface during the emotion stimuli. Evaporation of watery fluid results in abolish
12 Artificial Intelligence-Based Infrared Thermal Image Processing

of excess heat from the skin surface. Hence, thermal imaging technology is con-
sidered as a potential tool in evaluating the variation in cutaneous temperature and
its associated emotional states. The regions which are mostly produced a better
thermal response to emotional stimuli are the forehead, nose, palm, and maxillary
areas. Biological mechanisms such as emotional sweating and subcutaneous
vasoconstriction prevent blood flow to the skin surface. This occurs due to differ-
ent emotions such as pain, fear, and stress, and causes a reduction in temperature
in various ROIs. Besides other diagnostic modalities such as GSR, a thermal cam-
era non-invasively picturizing the activity of the sweat gland in the presence of
cold dots over the skin surface.
Koroteeva and Bashkatov (2021) were involved in studying the thermal signa-
ture related to the excretion of sweat droplet from the sweat pore and its evapora-
tion on the skin surface. Along with the thermal imaging, EEG and GSR were
measured using the multi-channel encephalography system. They produced a pre-
dominant sweat response from the study participants by means of applying various
external stimuli such as sudden mild electric shock, arithmetic mental task, or loud
sound to trigger emotional sweating. The thermal videos captured from the facial
regions are pre-processed and extracted using FLIR Research IR Max software. As
a first step, thermal videos are converted into frames and in each frame, the ROI is
fixed on the facial region containing the most predominant active sweat pores.
Each ROI contains a single sweat pore. Similarly, a number of ROIs are positioned
in every thermal frame. As the second step, an automated algorithm is used to com-
pute the sweat droplet progression using Matlab software. The variation in the
sweat droplet with time was measured based on the total number of pixels below a
certain threshold and validated with the manual method on the data set.
Further, the authors performed histogram analysis on the highlighted forehead
region of the face. They observed that arithmetic exercise provided as mental stim-
uli to the participants produces significant changes in the sweat gland activity. Due
to this, they detected the thermal profiles along the centerline of a single sweat
droplet. Similarly, they studied the activity of sweat glands in fingertips with the
onset of sudden loud sounds as stimulus. Based on the thermal imaging of the fin-
gertip, the diameter of the sweat droplet on the fingers is less than 1 mm. This size
is proportionate to the distance between the two neighboring papillary lines. They
observed that the palmar sweat gland has less secretion capacity than the face drop-
lets. They found minimal temperature difference (0.7 K) between the sweat pore
and surrounding skin in both face and fingers. Finally, they concluded that emo-
tional sweating due to functioning sweat glands on the face and fingers varies from
person to person. Hence, the thermal mapping of active cutaneous sweat glands for
each individual person is unique and can be used for biometric identification.
Krzywicki et al. (2014) developed a method for the identification of active
sweat glands and monitoring pore activity based on high spatial resolution ther-
mal imaging on finger and face regions. They demonstrated good association
between the skin conductance and activity of sweat glands obtained in thermal
imaging of eccrine glands in the forehead region. They quantified the pore activity
with high spatial resolution in the palm surface of medial, distal, middle, and
Fundamentals of Infrared Thermal Imaging 13

index fingers of the hand. This approach requires two infrared cameras equipped
with lenses of different focal lengths, and a specially designed finger-rest plate.
The pore activation index was calculated by means of counting the number of
active pores present in the skin area using thermal imaging. They implemented an
automated algorithm for counting the active pores using matched filtering tech-
niques and compared it with the manual method. From the thermal image of the
active pore, a pore template (7 × 7) is generated. Instead of applying smaller tem-
plates of size 5 × 5 or 3 × 3, a 7 × 7 pixel template was identified as the trade-off
between the computational efficiency and accuracy of the system. To identify the
active pores in the thermal image, the surrounding outer edges of 4 ROIs such as
index finger, medial–index finger, medial-middle finger and distal–middle finger
were defined manually. The normalized cross-correlation was applied between the
pore template and thermal image. If correlation values above a threshold of 0.6 are
reached, the PAI is recorded from each thermal video frame. Similarly, PAI was
obtained from the facial thermograms. A customized mask was manually designed
for each subject. The correlation-based tracking algorithm was used to monitor
the skin area. Template matching techniques are used to find the pixel closer to the
center of the pore than its edges. If the neighboring pixels near the pore center
exceed the correlation values above the threshold, sweat beads are measured.
The automated algorithm is evaluated and compared with the manual method
based on the pore count and pore activity on the thumb imaged by thermal video
segment by two independent evaluators Thermal videos of active pore regions
were captured for 30 s duration by two independent evaluators. Then, the thermal
videos are sampled at 25 Hz (750 frames) Considering the sluggish nature of pore
responses, the segments were down-sampled to 75 frames to achieve an efficient
result at the frame rate of 2.5 Hz. The activation response pores are clearly detect-
able from the segmented frame of the thermal video. To ensure the accuracy of the
test, thermal video segment was independently evaluated by two different evalua-
tors. It was evident that the evaluators are well versed with the qualitative thermal
pattern amalgamated with pores of thermal imaging.
The thermal image of middle and index fingers of subject after respiratory work
outs indicates the identification of thermal characteristics of the pores. A single
frame of each participant helps in providing the representative pore data. The cen-
ter location of each frame is highly considered as a reference to identify the thermal
characteristics of each pore. To identify the center location of a pore, the geometric
center of a pore was obtained from the pore activation index. Then, a 7 × 7 window
was placed over the geometric center. To fix the exact center point, the temperature
and correlation values calculated from the correlation map were attained for each
pixel within the 7 × 7 window. The temperature value present in the window were
transposed and rescaled so that minimum temperature mapped to “1” and maxi-
mum temperature mapped to “0.” The correlation coefficients for each pixel in 7 ×
7 window are multiplied with the newly mapped temperature to have the combined
pixel value of correlation and temperature. From this, the highest pixel value was
identified as the center pore location. The authors found a moderate relationship
between all PAI measures on the plantar side of the middle and index fingers and
14 Artificial Intelligence-Based Infrared Thermal Image Processing

skin conductance measured from index and middle fingers of the opposite hand.
However, after excluding the results of all low skin conductance responders and
considering only last three PAI measurements for correlations, a strong correlation
between PAI and SC was obtained. Compared to facial regions, thermal images of
fingers provided better information upon sweat gland activation.
Giacinto Di et al. (2014) investigated fear conditioning in post-traumatic stress
disorder (PTSD) using a standard GSR method and IR thermal imaging. The
authors obtained the gray scale digital images of different emotions such as anger,
happiness, and neutral. The participants were requested to count the number of
triangles and circles that were projected as images in front of the screen with sud-
den stimuli, like white noise burst, applied to them. The thermal video images
were captured using digital thermal camera FLIR SC 3000 FLIR Systems,
Sweden. Then, the sampling process was carried out by setting 10 frames/s. The
facial images of the participants were examined qualitatively for their changes in
autonomic responses by means of visual inspection. Then, the quantitative analy-
sis of temperature changes in the cutaneous region of the facial thermal image was
evaluated using Matlab software. The circular ROI is fixed in the nose region of
the thermal image and its corresponding thermal profiles were obtained. The dis-
placement between the ROI locations in successive thermal image frames is cor-
rected by using motion correction and image registration techniques to avoid
mismatch in fixing of ROIs. The image registration process involves re-alignment
of ROI by using automated recognition and identification of anatomical regions of
the nose. After the acoustic stimulation, a decrease in temperature was observed
in the entire face due to sudomotor response. The temperature pattern correspon-
dence to emotional sweating response shows the colder dotted spots in nose tip,
maxillary, and forehead regions. They found that skin conductance response and
facial thermal response provided good investigation methods in the psychometric
analysis of PTSD conditions.

1.3.4 Facial Muscle Activation Signals

Muscle activation in the face may be indicated by thermal imaging. Human
facial expressions result in the activation of muscles which may be associated
with changes in skin temperature distribution. The facial action coding system
(FACS) and facial surface EMG are the two different methods used to study the
characteristics of facial muscle activations. FACS disintegrates facial expressions
into discrete action units and combines them together to form more elaborate
expressions.
Jarlier et al. (2011) conducted the experiment to study the specific facial muscle
heat patterns for examining the sensitivity of thermal imaging. Their ultimate aim
is to identify the facial region and perform face normalization followed by feature
extraction and facial expression classification based on facial features. During the
image acquisition process, the participant’s heads were focused toward the imaging
system in immobilized condition with a head fixation system. The participants
were informed to perform five action units with different intensity–speed
Fundamentals of Infrared Thermal Imaging 15

combinations. Initially, the acquired images were resized to the facial area alone by
using the bilinear interpolation method. The images are aligned with each other
using 2D translation and rotation procedures. The images are realigned using an
optimization algorithm to obtain the transformation that restores the original image.
The affine transformation method is used to match the control point marked in the
individual face with the average facial image present in the database. Then, the
authors used spatial patterning approach to detect the changes in temperature pat-
terns. They applied spatial PCA to determine the relationship between the tempera-
ture values measured for each pixel in the facial regions. Spatial PCA was performed
in each trial for participants, considering the pixel as a variable and temperature
value as observation. The PCA finds the complex relationship between the tem-
perature values measured for each pixel. The result obtained from PCA is a set of
factor loading which corresponds to the spatial component of the original tempera-
ture values. The resultant PCA consists of a set of factor loadings from which the
first two components were taken into consideration and averaged to produce the
facial areas where temperature values depend on action unit performance.
After finding the principle components, the peak temperature (apex) from the
facial region is determined. To measure the maximum temperature variations, an
automated baseline to peak analysis is performed in order to find the apex value.
The speed of the temperature variations has been determined from the latency of
the apex. The intensity of temperature variations was obtained from the signal
amplitude at the apex. The features extracted from the thermal profiles are a repre-
sentative temperature map and a mean representative topographic map. The repre-
sentative temperature map was obtained by subtraction of the temperature value
measured at the apex with the baseline temperature. The average representative
topographic map is calculated for each action unit by averaging all the individual
maps attained in each condition. The quantification of thermal imaging parameters
related to facial muscle contractions was classified using the k-nearest neighbor
model. The representative temperature map parameter and average representative
topographic parameter were given as the input to the classifier to classify the dif-
ferent emotions based on the thermal pattern of facial muscle activity.
Wesley et al. (2012) performed automated facial recognition by comparing the
video of thermal imaging, digital facial videos, and fusion of thermal and digital
facial images. In their study, the subjects were trained by using a FACS encoder
playing the videos of facial expressions and provided sufficient time for practicing
each facial expression. Initially, neutral and relaxed expression is captured in both
thermal camera and digital camera. Then, the participants were informed to repeat
each facial expression 14 times and recorded to simulate various natural expres-
sions. The authors used local feature extraction method in which only the facial
areas which are subject to change with facial expressions are considered for pro-
cessing. To extract the features, ROIs are fixed at 13 fiducial points on the facial
region such as forehead, forehead bruise, below the eyes, left and right chin, and
cheek. The mean temperature value was computed in each ROI in neural state for
the first 25 s. Then, the principle components were extracted from each ROI by
considering each pixel value within the ROI as variable. The highest principle
16 Artificial Intelligence-Based Infrared Thermal Image Processing

component values are obtained from the frames associated with the largest varia-
tion from the neutral ROI. After the determination of principle components from
all ROIs, a profile for each facial expression is obtained by finding the standard
deviation of each ROI. These facial expressions indicate the deviation over the
course of expression. This standard deviation obtained from the profile was used
for training the feed-forward multi-layer perceptrons for both thermal and visual
images. A decision-level fusion method is used for the fusion of thermal and
visual facial images. From their results, it was observed that visual facial images
produced better performance only under the constant lighting conditions, but ther-
mal imaging modality produced good results even if there is heat variation in the
dataset. However, the fusion of thermal and visual imaging modalities has pro-
duced better performance compared to individual imaging modalities.

1.4 EXAMPLES OF RECENT APPLICATIONS OF INFRARED

THERMAL IMAGING IN MEDICINE
The following section summarizes articles reporting thermographic studies
employing modern infrared imagers and artificial intelligence methods to detect
joint inflammation, skin diseases, dental or metabolic disorders, or as a method
for fever screening. The camera specifications employed in each application are
collected in a table, and the applied method of image processing is also described.
Finally, the clinical value of each study is critically discussed with respect to the
performance of thermal imaging as a diagnostic tool or as an outcome measure.

1.4.1 Rheumatoid Arthritis
Infrared thermal imaging has been used for more than five decades for the assess-
ment of inflammatory disorders in various small and large joints of the human
body. In 1975, Francis Ring reported in his [Link]. thesis the fundamentals for
applying the technique in rheumatology research and practice (Ring, 1975).
Michael Engel published in 1984, reference values for unaffected and inflamed
joints including the wrist, metacarpophalangeal joints, proximal interphalangeal
joints, elbow, knees, ankles, metatarsophalangeal joints, and sternoclavicular and
ileosacral joints (Engel and Saier, 1984). The early thermographic studies in rheu-
matology suffer from the fact that thermal findings were seldom related to clinical
signs of inflammatory arthritis, although first disease activity scores such as the
Lansbury index were already available.
Pauk et al. (2019a) investigated the hands of RA patients with high and moder-
ate disease activity in comparison to healthy controls using infrared thermal cam-
era FLIR E60bx with a sensitivity of 0.405°C and an accuracy of less than 2% and
thermal resolution of 0.03 K. The authors recorded the thermograms in the finger
region in three different stages: static temperature, thermal video recordings dur-
ing cooling, and recording during the rewarming process. The authors performed
image pre-processing operations such as normalization and filtering. Then, they
applied morphological erosion and dilation to extract the finger boundaries. Later,
a skeletonization algorithm and a modified depth-first search algorithm were
Fundamentals of Infrared Thermal Imaging 17

applied to extract the required ROI and to remove the outer layers of the hand
image. The resulting ROI is the one-pixel wide middle line in each finger. They
obtained different temperature findings between the three groups investigated.
However, the study design missed detecting classical signs of joint inflammation
such as swelling (tumor), redness (rubor), and warmth (calor), which can easily be
captured by inspection and palpation. In addition, extending the linear ROI to the
fingertips beyond the distal interphalangeal joint, which is not affected by rheu-
matoid arthritis, may have resulted in the overestimation of finger temperature in
healthy subjects, who present typically with a fingertip temperature ± 0.5°C of the
metacarpophalangeal temperature. The difference between maximum and mini-
mum temperatures obtained in the finger (Pauk et al. 2019b) was used to compress
the temperature information of the linear ROI into a single value but it cannot
localize the site of maximal temperature.
It is essential for automatic generation of ROI that they are placed at a mean-
ingful anatomical region. In case of rheumatoid arthritis hands, the recommended
positions are the wrist, and the metacarpophalangeal and proximal interphalan-
geal joints 1 (thumb) to 5 (little finger). Such an array of ROI would allow inves-
tigating the coincidence of hyperthermic ROIs with tender or swollen joints, since
this information is already available in the individual records of the number and
distribution of affected joints. Although time-consuming, manual delineation of
ROIs may manage this task better than the automatic generation of measurement
areas. The number of hyperthermic joints might become an alternative to the num-
ber of swollen or tender joints for the evaluation of disease activity. Unfortunately,
data supporting this approach are yet not available.
Snekhalatha et al. (2015, 2018) evaluated the RA using thermal imaging of the
hand region. They used hand-held thermal camera ThermaCAM T400 with a resolu-
tion of 320 × 420 and a temperature sensitivity of <50 mK at 30°C. They implemented
k-means and fuzzy c-means algorithm for the segmentation of hand region in RA and
normal subjects. They extracted the statistical features using gray level co-occurrence
matrix algorithm. The authors developed computer-aided diagnostic tool for auto-
mated classification of images from RA patients or healthy subjects. They predicted
that the performance of k-means algorithm was better compared to fuzzy c-means in
the segmentation of hand thermal images for the total studied population. They found
a significant correlation between the skin temperature measured at 2nd and 3rd MCP
finger joints with features extracted from segmented hand regions. The main limita-
tion of this study is the lack of coincidence of temperature increase beyond a defined
threshold with clinical features such as joint swelling or tenderness. The Second and
third MCP are frequently affected by RA and the correlation between extracted fea-
tures and temperature at this site was found to be significant, however, the agreement
of these findings with clinical signs must be questioned. Particularly, the example of
the thermal image of a rheumatoid arthritis with homogeneously distributed tempera-
tures between 35.4 and 35.8°C on the hand from the interdigital skinfold up to the
distal forearm is not a typical finding in rheumatoid arthritis where a focal hyperther-
mia above the inflamed joints would be expected.
Gatt et al. (2020) studied the thermal characteristics of RA feet during the
clinical remission stage. The authors acquired the thermal images in the foot
18 Artificial Intelligence-Based Infrared Thermal Image Processing

region using a FLIR T630 camera. The thermal camera has a focal plane array
with a resolution of 640 × 480 pixels and a sensitivity of <40 mK. They obtained
the mean temperatures in the following ROI such as medial, lateral, and central
feet and heel regions. The authors found a statistically significant difference
between RA and normal subjects in all the forefeet regions. Like in hands, how-
ever, focal hyperthermia is only expected in skin regions in close vicinity to
inflamed joints, projecting to the distal medial and lateral foot region, and the
observed increase in foot sole temperature may not be necessarily caused by the
pathology involved in rheumatoid arthritis. Gatt et al. concluded that large, multi-
center randomized clinical trials are needed in order to determine the sensitivity
and specificity of this imaging modality to finally establish whether it is actually
valid and reliable, sufficient to be applied to clinical practice as a screening tool
for the timely detection of joint inflammations.

1.4.2 Diabetes Mellitus
Gururajaroa et al. (2019) conducted a study for the detection of foot ulcers in
diabetic neuropathy patients, which describes the conditions for developing soft-
ware to assess the foot temperatures and differentiates the feet of healthy subjects
from the that of ones with diabetes. The authors applied active contour deform-
able model in combination with morphological processing for the segmentation of
foot of patients with diabetes. They extracted the ROI from the segmented image
and used the bounding box method to partition the left and right foot regions.
They extracted the features and performed asymmetry analysis to compare the left
and right foot regions. They also implemented a pre-trained CNN called Mobile
net for the classification of normal, diabetes with, and without complications.
However, clinical features of healthy controls, diabetes, and diabetes complica-
tions are not reported. The significance of the detected variation in temperature of
the plantar foot between the investigated three groups remains unknown since the
investigated population was small despite data augmentation.
Liu et al. (2015) developed an imaging system composed of thermal and visual
color images that is capable to perform an automatic analysis of the plantar feet of
diabetes patients. They used a FLIR SC305 thermal camera with a resolution
320 × 240 pixels and an accuracy of ± 2°C. The authors performed foot segmenta-
tion in color images using k-means clustering algorithm. They also performed an
image registration process to align the segmented color image with the foot ther-
mal image. Based on asymmetry analysis, the system missed only 2 out of 37 foot
ulcers in the thermal images.

1.4.3 Dermatology
Cellulite is an alteration of the topography of the skin that occurs mainly in
women in the pelvic region, lower limbs, and abdomen. It is characterized by
a padded or “orange peel” appearance (Rossi and Vergnanini, 2000). Bauer et
al. (2018) conducted a feasibility study on the utility of infrared thermography
Fundamentals of Infrared Thermal Imaging 19

for the classification of cellulite stages. They used thermal camera FLIR T335
which has a resolution of 320 × 240 pixels and a sensitivity of 50 mK. They
automatically adjusted all recorded thermal images to the same temperature win-
dow. They found that cellulite regions exhibit non-uniform temperature distribu-
tion in thigh and lower back regions characterized by contours and shapes with
uneven and higher temperatures than the surroundings. Those irregularities were
marked manually using Image J software. They extracted four characteristics
from the thermal images, such as Group 1: Number of irregularities; Group 2:
Cumulative area; Group3: Cumulative area of irregularities/area of thigh region;
Group4: Irregularities number × area of irregularities. These features were used
in a deep learning algorithm to differentiate cellulite stages, which were after
further modification of the classifiers able to detect the grade of cellulite cor-
rectly in 48 out of 49 thermal images. Unfortunately, scoring the clinical assess-
ment of cellulite patients is incompletely reported, particularly the number of
false-positive and false-negative cases of the thermographic evaluation remains
obscured. Nevertheless, this promising imaging-orientated approach warrants fur-
ther investigations.
Herman and Cetingul (2011) acquired images of pigmented lesions with both
digital and thermal cameras using a square adhesive marker for the identification
of the lesion’s location. A landmark detection algorithm was used to identify the
corners of the square marker. Further image processing included a random walker
segmentation algorithm to create a mask image for the identification of skin
lesions and a quadratic motion model to compensate motion-induced artifacts.
Mild convective cooling was initiated after the baseline images have been cap-
tured. The temperature recovery was observed, and different rewarming patterns
were reported for healthy skin and lesioned skin presenting with signs of malig-
nancy in the histological evaluation of tissue obtained from biopsy.

1.4.4 Dental Disorders
Ammoush et al. (2018) conducted the study to evaluate the role of thermography
in the differentiation of patients with suspected dental abscess or facial cellulitis.
The authors used a FLIR C2 infrared camera to acquire the thermal image of
both lateral views of the face. They obtained higher temperature differences of
2.4°C in facial cellulitis patients compared to dental abscess patients whose tem-
perature difference was 1.5°C between the normal site and the affected site. The
absence of criteria for diagnosing cellulitis is the main weakness of this study, but
the association of increased temperature in the skin overlying suspected odon-
togenic inflammation warrants further investigation. Macianskyte et al. (2019)
examined the relationship between infrared thermography and computed tomog-
raphy results by comparing asymmetrical temperature distribution obtained from
infrared thermal imaging with the CT lesions. Initially, they detected the human
face edges using image gradients and convolution max. They identified the facial
symmetry axis to separate the left and right facial regions and calculated the aver-
age mean temperature on both the sides of facial regions. They performed image
20 Artificial Intelligence-Based Infrared Thermal Image Processing

segmentation of left and right ROI using the binary thresholding method based
on neighboring temperatures. They obtained a higher average temperature differ-
ence in the facial region compared to the mouth region in tumor patients. They
observed a negligible temperature difference in the facial region than the mouth
region in healthy subjects. The authors predicted that the asymmetrical tempera-
ture zone detected from thermal images matches or aligns with the presence of
maxilla-facial pathologies obtained from CT images. However, some maximal
temperatures at the lesion site were 0.8–2.0°C higher than the core temperature of
37.0°C, questioning the accuracy of temperature readings extracted from the ther-
mal images. In addition, the evaluation of the diagnostic accuracy of temperature
asymmetry for the detection of tumorous lesions visible in CT images is unclear
in terms of thresholds for both temperatures and spatial distribution. These find-
ings are in contradiction to both the thermodynamics and thermal physiology of
heat transfer to the surface.
Kaspryzk et al. (2021) observed the temperature course in buccal skin before,
immediately, 1 and 7 days after surgical extraction of a retained third molar. They
carried out the thermal imaging with the head positioned in the sagittal plane
using thermal camera FLIR E60. They defined two circular ROIs, a large one
including the total bucca, and a small one at the projection site of the third molar.
A decrease in temperature of 1.5°C was observed from day 1 to day 4 and a sud-
den increase in temperature occurred from day 4 to day 7. Unfortunately, the
authors did not report the findings of clinical inspection of the wound. Thus, their
interpretation of skin temperature changes is completely based on the typical his-
tological findings during wound healing. Relating the initial drop in temperature
to “repair processes” and the following temperature increase to granular tissue
formation is not yet confirmed by controlled studies. Finally, the authors expressed
their opinion that thermal imaging might be helpful in monitoring patients after
tooth extraction.
Iosif et al. (2016) investigated candida-associated denture stomatitis using
infrared thermography based on the comparison of normal subjects and a dis-
eased group. The authors used a ThermaCAM PM 350 infrared camera to acquire
thermal images from the maxillary denture bearing area. The authors found sta-
tistically significant higher temperatures in candida-associated denture stomatitis
(36.20°C) compared to the healthy oral mucosa (34.85°C). The authors reported
an optimal thermal threshold of 35.4°C for differentiating normal oral mucosa
from the mucosa in candida-associated denture stomatitis. Finally, they con-
cluded that infrared thermography can be used as a complementary investigative
method to diagnose inflammatory mouth disorders such as candida-associated
denture stomatitis.
Haddad et al. (2016) proposed reference values for normal temperatures at spe-
cific sites of the face. The authors used ThermaCAM T400 FLIR systems to acquire
the facial thermal image in frontal, lateral, and medial views. They captured thermal
images with a size of 10x10 wide window and identified the hottest areas on the
forehead, around the eyes and the nose, and in the vicinity of the mouth. Then, they
compressed the temperature window by 50% and selected the unchanged hot areas
Fundamentals of Infrared Thermal Imaging 21

as the site of temperature reference. A strange definition of thermal gradient was

used as rationale for selecting the measurement sites. Since a thermal gradient is
defined by the temperature difference along a distance, the mean or maximum tem-
perature obtained from a small circular measurement area does not describe a ther-
mal gradient. In fact, any non-homogenous temperature field can be described by
multiple thermal gradients, orientated into multiple directions. Thus, for defining
the thermal gradient the boundaries of the corresponding measurement area must be
defined. The authors defined the ROI by the arteries and veins providing perfusion
of the region where the reference is located and by the sensory innervation field of
terminal branches of the trigeminal nerve. However, clear boundaries of ROIs are
not defined and the thermal gradient, equal to the difference between minimal and
maximal temperature within the ROI, remains undefined.
Only the images of 50 subjects without symptoms or history of a cranio-facial
disorder were analyzed for normal reference values. Receiver operator character-
istics (ROC) analysis was obtained in which temperature dispersion parameter
outperformed the best in detecting any reference site. At thresholds of 34.6°C in
the frontal view and 34.8°C in lateral views, mean and maximum temperatures
performed best with an area under the curve (AUC) of 0.839 for mean and 0.876
for maximum temperature.
In a larger sample, also including 111 subjects with minor symptoms, the side-
to-side difference in all 14 pairs of reference sites was in the range between 0.07
and 0.52°C, with minimal differences at the temples and maximal differences at
the nasolabial folds. Men presented in each reference site with slightly higher
temperatures than women, but at the nasolabial folds, the mean difference was
large: 1.1°C.

1.4.5 Complex Regional Pain Syndrome

Complex regional pain syndrome (CRPS) is a chronic pain syndrome which affects
the upper and lower extremities following injuries, often of minor in nature, but
surgery or in rare cases even heart infarction or a lung tumor may be the initial
trigger. Early diagnosis and treatment of CRPS improve the functional prognosis.
The International Association for the Study of Pain (IASP) classified CRPS into
two types as follows: Type I, formerly known as “reflex sympathetic dystrophy,”
in which a peripheral nerve injury is not observed, and type II, formerly known
as “causalgia.” Diagnosis of CRPS is based on the “Budapest Criteria,” where a
temperature difference of 1°C between affected and non-affected limb is required
to evidence vasomotor disturbance. There are several non-invasive methods avail-
able to detect the components of CRPS. The skin temperature changes can be
measured by a thermal camera and a GSR sensor may be used for sweating detec-
tion. Quantitative sensory testing identifies sensory deficits, trophic alterations,
and edematous swellings found by inspection and palpation. Pain related informa-
tion is captured in the patient’s history and quantified by a pain scale.
Niehof et al. (2006) conducted a study using thermography to evaluate the
response to whole-body cooling followed by whole-body warming in patients
22 Artificial Intelligence-Based Infrared Thermal Image Processing

with unilateral complex regional pain syndrome at the hands in comparison to

healthy controls. The authors used a ThermaCAM SC2000 FLIR thermal camera
to measure the skin temperature of all fingertips on the dorsal side of both hands.
A visual analog scale (VAS) was used to record the pain in the hand region simul-
taneously with thermal imaging. The authors interpreted the difference in finger-
tip temperature between the affected and the non-affected hand in patients and
between the dominant and the non-dominant hand in controls as a marker of the
vasoconstrictor activity. Also, the differences between the minimal fingertip tem-
perature and the maximal fingertip temperature during the whole temperature
cycle were calculated in both the involved and contralateral hands. The asymme-
try factor, based on histograms of temperature distribution was also used for eval-
uation. At baseline, the median difference in fingertip temperature was 0.43°C
(0.04–0.66°C) in controls and 0.37°C (0.1–0.77°C) in patients. During the total
temperature cycle, the average fingertip temperature in controls was a median of
0.95°C (0.50–1.51°C) and in patients, the median was 2.50°C (1.61–3.43°C).
Hence, CRPS patients present with a wider response of skin temperature to whole-
body thermal stimulation than healthy controls. The best performance as a diag-
nostic test for CRPS was obtained for the minimum asymmetry factor during the
whole temperature cycle with a diagnostic sensitivity of 1 and a specificity of
0.83, followed by the asymmetry factor at rest (sensitivity 1, specificity 0.75).
The maximum average fingertip temperature difference during temperature cycle
(sensitivity 0.92, specificity 0.75) and the average fingertip temperature difference
at rest (sensitivity 0.76, specificity 0.38).
Cho et al. (2016) performed a retrospective study to evaluate the temperature
difference distribution between the affected and unaffected limbs in CRPS
patients. They observed about 44.3% of patients presenting with 1°C temperature
difference between the bilateral limbs. This is not an unexpected finding, because
the established diagnostic Budapest criteria, allow the diagnosis of CRPS when
the “vasomotor” category is explained by color changes and clinical signs are
available in at least two of the four diagnostic categories sensitivity, vasomotor,
trophic/motor and sudomotor/swelling. Since temperature was not a predominant
sign of CRPS in the investigated population, absence of significant correlations
between skin temperature and symptom duration is not surprising.
Radius fracture is a frequent initial injury triggering the development of CRPS.
Ammer (1991a), conducted a study on patients, whose plaster fixation was
removed due to radiographic signs of bone healing (Ammer, 1991a). Thermal
images were recorded immediately after removal of the fixation and a week later,
radiography of the distal forearm and the hand skeleton was performed.
Radiographs were evaluated for early signs of algodystrophy, which were found
in 21 of 41 patients. Mean skin temperature in an area of the wrist and the hand
dorsum was based on 10 spot temperatures located at the wrist and the midpoint
of each metacarpal bone. Side-to-side temperature difference (affected side minus
non-affected side) was calculated. At baseline, the hand temperature on the injury
side was 32.8 ± 1.4°C and 31.9 ± 1.7°C on the control side. A week later, the mean
temperature of the injured arm decreased to 32.6 ± 1.2°C, while the non-injured
Fundamentals of Infrared Thermal Imaging
Ammer K Thermographie nach gipsfixierter Radiusfraktur. Thermologie Österreich 1991a,
1(1): 4–8.
Ammer K Thermographische Therapieüberwachung bei [Link]. ThermoMed 1991b, 7:
112–115.
Ammer K and Ring G. The Thermal Human Body. A Practical Guide to Thermal Imaging.
Jenny Standord Publishing, Singapore, 2019.
Ammoush M , Gzawi M , Warawreh A , Hijazin R , Jafar H. Clinical evaluation of
thermography as a diagnostic tool in oral and maxilla-facial lesions. Journal of Royal Medical
Services 2018, 25: 45–49.
Anbar M Recent technological developments in thermology and their impact on clinical
applications. Biomedical Thermology 1990, 10(4): 270–276.
Anbar M Dynamic thermal assessment. In: Medical Infrared Imaging: Principles and
Practices; edited by Mary Diakides , Joseph D. Bronzino and Donald R. Peterson , CRC
Press, 2013, pp. 8.1–8.23.
Baic A , Kasprzyk T , Rżany M , Stanek A , Sieroń K , Suszyński K , Marcol W , and
Cholewka A. Can we use thermal imaging to evaluate the effects of carpal tunnel syndrome
surgical decompression? Medicine (Baltimore) 2017, 96(39): e7982.
Bauer J , Grabarek M , Migasiewicz A , Podbielska H. Non-contact thermal imagigng as
potential tool for personalized diagnosis and prevention of cellulite. Journal of Thermal
Analysis and Calorimetry 2018, 133: 571–578.
Becker N Screening aus epidemiologischer Sicht. Radiologe 2002, 42: 592–600.
Biró TS How to measure the temperature. In: Is There a Temperature? Fundamental
Theories of Physics, (2011) vol 171. Springer, New York, NY. doi: 10.1007/978-1-4419-8041-
0_2
Chekmenev SY , Rara H , Farag AA . Non-contact wavelet-based measurement of vital signs
using thermal imaging In: The First International Conference on Graphics, Vision, and Image
Processing (GVIP), Cairo, Egypt, 2005, pp. 107–112.
Cho W , Nahm FS , Choi E , Lee PB , Jang IK , Lee CJ , Kim YC , Lee SC . Multi-center
study on the asymmetry of skin temperature in complex regional pain syndrome. An
examination of temperature distribution and symptom duration. Medicine 2016, 95(52):
e5548.
Clay-Warner J and Robinson DT Infrared thermography as a measure of emotion response.
Emotion Review 2015, 7(2): 157–162.
Cobet R Die Hauttemperatur des Menschen. Ergebnisse der Physiologie 1926, 25(1):
439–516.
Conwell TD , Hobbins WB , Giordano J. Sensitivity, specificity and predictive value of infrared
cold water autonomic functional stress testing as compared with modified IASP criteria for
CRPS. Thermology International 2010, 20(2): 60–68.
Conwell TD , Lind KE . Comparison of the diagnostic accuracy of three infrared imaging
methods in evaluating patients with presumptive complex regional pain syndrome, TypeI.
Thermology International 2015, 25(2): 54–63.
Engel J-M , Saier U Thermographische Standarduntersuchungen in der Rheumatologie und
Richtlinien zu deren Befundung. Luitpold, München, 1984.
Fei J , Pavlidis I , Thermistor at a distance: Unobtrusive measurement of breathing. IEEE
Transactions on Biomedical Engineering 2010, 988–998.
Garbey M , Sun N , Merla A , Pavlidis I. Contact free measurement of cardiac pulse based on
the analysis of thermal imagery. IEEE Transaction on Biomedical Engineering 2007,
1418–1426.
Gatt A , Mercieca C , Borg A , Grech A , Camilleri L , Gatt C , Chockalingam N , Formosa C.
Thermal characteristics of rheumatoid feet in remission: Baseline data. Plos One 2020,
15(12): e0243078.
Giacinto Di A , Brunetti M , Sepede G , Ferretti A , Merla A Thermal signature of fear
conditioning in mild posttraumatic stress disorder. Neuroscience 2014, 266: 216–223.
Gururajaroa SB , Venkatappa U , Shivaram J M , Sikkandar MY , Amoudi AA . Infrared
thermography and soft computing for diabetic foot assessment. Machine Learning in
Biosignal Analysis and Diagnostic Imaging 2019; 73–97.
Haddad DS , Brioschi ML , Baladi MG , Arita ES . A new evaluation of heat distribution on
facial skin surface by infrared thermography. Dentomaxillo Facial Radiology 2016, 45:
20150264.
Herman C , Cetingul MP . Quantitative Visualization and detection of skin cancer using
dynamic thermal imaging. The Journal of Visualized Experiments 2011, 51: e2679.
Houdas Y , Ring EFJ Human Body Temperature: Its Measurement and Regulation. Plenum
Press, New York, London, 1982.
Huygen FJ , Niehof S , Klein J , Zijlstra FJ Computer-assisted skin video thermography is a
highly sensitive quality tool in the diagnosis and monitoring of complex regional pain
syndrome type I. European Journal of Applied Physiology 2004, 91(5): 516–524.
Ioannou S , Gallese V , Merla A. Thermal infrared imaging in psychophysiology: Potentialities
and limits. Psychophysiology 2014, 51: 951–963.
Iosif L , Proteasa CT , Magureanu CM , Proteasa E. Clinical study on thermography as
modern investigation method for candida-associated denture stomatitis. Romanian Journal of
Morphology and Embryology 2016, 57: 191–195.
Jakkaw P , Onoye T. Non-contact respiration monitoring and body movements detection for
sleep using thermal imaging. Sensors 2020, 20: 6307.
Jarlier S , Grandjean D , Delphanque S , Diaye KN , Cayeux I , Velazco MI et al. Thermal
analysis of facial muscles contractions. IEEE Transactions on Affective Computing 2011,
2(1): 2–9.
Kaspryzk KT , Cholewka A , Balamut K , Kownacki P , Kaszuba N , Kaszuba M , Staneck A ,
Sieron K , Stransky J , Pasz A , Morawiec T. The applications of infrared thermography in
surgical removal of retained teeth effects assessment. Journal of Thermal Analysis and
Calorimetrey 2021, 144: 139–144.
Kim Y , Park Y , Kim J , Lee EC . Remote heart rate monitoring method using Infrared
thermal camera. International Journal of Engg Research and Technology 2018, 11(3):
493–500.
Koroteeva EY , Bashkatov AA . Thermal signatures of liquid droplets on a skin induced by
emotional sweating QIRT. Quantitative InfraRed Thermography Journal 2021, 19(2):
115–125.
Krzywicki AT , Berntson GG , Kane BLO. A non-contact technique for measuring eccrine
sweat gland activity using passive thermal Imaging. International Journal of
Psychophysiology 2014, 94(1): 25–34.
Lewis GF , Gatto RG , Porges SW . A novel method for extracting respiratory rate and
relative tidal volume from infrared thermography. Psychophysiology 2011, 48(7): 877–887.
Liu C , Van Netten JJV , Van Baal JG , Bus SA , Vander Heijde F. Automatic detection of
diabetic foot complications with infrared thermography by asymmetric analysis. Journal of
Biomedical Optics 2015, 20(2): 026003.
Machin G , Tsai BK . Temperature Fundamentals. In: Radiometric Temperature
Measurements, edited by Z. M. Zhang , B. K. Tsai and G. Machin , Academic Press, Oxford,
UK, 2010, pp. 29–71.
Macianskyte D , Monastyreckiene E , Basevicius A , Adaskervicius R. Comparison of
segmented thermal images versus a CT scanning for detection of maxillofacial pathology.
Dentomaxillofacial Radiology 2019, 48: 20180075.
Nguyen AV , Cohen NJ , Lipman H , Brown CM , Molinari NA , Jackson W , Kirking H et al.
Comparison of 3 infrared thermal detection systems and self-report for Mass fever screening.
Emerging Infectious Diseases 2010, 16: 1710–1717.
Niehof S , Huygen FJPM , Vander Weerd RWP , Westra M , Zijlstra FJ . Thermography
imaging during static and controlled thermoregulation in complex regional pain syndrome
type I: Diagnostic value and involvement of the central sympathetic system. Biomedical
Engineering Online 2006, 5: 30.
Niehof SP , Beerthuizen A , Huygen FJ , Zijlstra FJ Using skin surface temperature to
differentiate between complex regional pain syndrome type 1 patients after a fracture and
control patients with various complaints after a fracture. Anesthesia & Analgesia 2008,
106(1): 270–277.
Pauk J , Ihnatouski M , Wasilewska A Detection of inflammation from finger temperature
profile in rheumatoid arthritis. Medical and Biological Engineering and Computing 2019a,
57(12): 2629–2639.
Pauk J , Wasilewska A , Innatouski M Infrared thermography sensor for disease activity
detection in Rheumatoid arthritis patients. Sensors 2019b, 19: 3444.
Pavlidis I , Tsiamyrtzis P , Shastri D , Wesley A , Zhou Y , Linder P , Buddharaju P , Joseph
R , Mandapati A , Dunkin B , Bass B. Fast by nature-How stress patterns define human
experience and performance in dexterous tasks. Scientific Reports 2012, 2: 305.
Pereira CB , Czaplik M , Blazek V , Leonhardt, Teichmann D. Monitoring of cardiorespiratory
signals using Thermal imaging: A pilot study on Healthy human subjects. Sensors (Basel)
2018, 18(5): 1541.
Perez-Concha T , Tijero B , Acera M , Fernanadez T , Gabilonda I , Gomez-Esteban JC .
Usefulness of thermography in the diagnosis and classification of complex regional pain
syndrome. Neurologia (Engl Ed). 2020 Dec 17:S0213-4853(20)30337-6. English, Spanish.
doi: 10.1016/[Link].2020.10.011. Epub ahead of print. PMID: 33342641.
Ring EFJ The Thermographic Assessment of Anti-Inflammatory Drug Therapy in Arthritis.
[Link]. Thesis, University of Bath, 1975.
Ring EFJ Technical advances in thermal imaging. In: Thermal Assessment of Breast Health,
edited by M. Gautherie , E. Albert and L. Keith , MTP Press, Lancaster. 1984, pp. 3–10.
Rossi ABR , Vergnanini AL Cellulite: A review. Journal of the European Academy of
Dermatology and Venereology 2000, 14(4): 251–262.
Schuhfried O , Herceg M , Reichel-Vacariu G , Paternostro-Sluga T. Infrared thermographic
pattern of lower limb complex regional pain syndrome (Type I) and its correlation with pain
disease duration and clinical signs. Physikalische Medizin, Rehabilitationsmedizin,
Kurortmedizin 2016, 26(6): 288–292.
Snekhalatha U , Anburajan M , Sowmiya V , Venkatraman B , Menaga M Automated hand
thermal image segmentation and feature extraction in evaluation of rheumatoid arthritis.
Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in
Medicine 2015, 229(4): 319–331.
Snekhalatha U , Somiya V , Nilkantha G Computer aided diagnosis-based hand thermal
image analysis a potential tool for the evaluation of Rheumatoid arthritis. Journal of Medical
and Biological Engineering 2018, 38: 666–677.
Wesley A , Buddharaju P , Pienta R , Pavlidis I. A Comparative Analysis of Thermal and
Visual Modalities for Automated Facial Expression Recognition. Advances in Visual
computing. Lecture notes in CS. Springer, Berlin Heidelberg 2012 pp. 51–60.
Zhang ZM , Machin G Overview of Radiation Thermometry. In: Radiometric Temperature
Measurements, edited by Z. M. Zhang , B. K. Tsai and G. Machin , Academic Press, Oxford,
UK, 2010, pp. 1–28.
Zhou Y , Ghassemi P , Chen M , McBride D , Casamento JP , Pfefer TJ , Wang Q. Clinical
evaluation of fever screening thermography impact of consensus guidelines and facial
measurement location. Journal of Biomedical Optics 2020, 25(9): 097002.

Protocol for Standardized Data Collection in Humans

Amalric R , Giraud D , Altschuler C , Deschanel J , Spitalier JM Analytical, synthetic, and
dynamic classification of mammary thermograms. Acta Thermographica 1978, 3: 5–17.
Amalric R , Giraud D , Altschuler C , Spitalier JM Value and interest of dynamic
telethermography in detection of breast cancer. Acta Thermographica 1976, 1: 89–96.
Ammer K Thermal evaluation of tennis elbow. In: Ammer K. , EFJ Ring (eds), The Thermal
Image in Medicine and Biology. Wien: Uhlen Verlag, 1995, pp. 214–219,
Ammer K Repeatability of temperature measurements at the forehead in thermal images from
the standard view “face” Thermology International 2006, 16(4): 138–142.
Ammer K The Glamorgan Protocol for recording and evaluation of thermal images of the
human body. Thermology International 2008a, 18: 125–144.
Ammer K Thermal imaging: A diagnostic aid for fibromyalgia? Thermology International
2008b, 18(1): 45–50.
Ammer K Cold challenge to provoke a vasospastic reaction in fingers determined by
temperature measurements: A systematic review. Thermology International 2009, 19:
109–118.
Ammer K Repeatability of identification of hot spots in thermal images is influenced by image
processing. Thermology International 2011, 21(2): 40–46.
Ammer K Temperature of the fingertips in subjects with suspected Raynaud’s phenomenon.
In: Nowakowski A , Mercer J. (eds) Lecture Notes of the ICB Seminar. Springer, Warsaw,
2013, 94, pp. 95–98.
Ammer K , Formenti D Does the type of skin temperature distribution matter? Thermology
International 2016, 26(2): 51–54.
Ammer K , Melnizky P , Schartelmüller T Thermographie bei Fazialisparese ThermoMed
1997, 13: 6–11.
Ammer K , Ring EF Repeatability of the standard view of both dorsal hands. Results from a
training course on medical infrared imaging. Thermology International 2004, 14(3): 100–103.
Ammer K , Ring EFJ. Standard procedures for infrared imaging in medicine. In: Diakides NA
and Bronzino JB (eds) Medical Infrared Imaging. Boca Raton: CRC Press 2008, 22, pp.
1–122.14.
Ammer K , Ring F The Thermal Human Body. A Practical Guide to Thermal Imaging.
Singapore: Jenny Stanford Publishing, 2019.
Bergersen TK A search for arteriovenous anastomoses in human skin using ultrasound
Doppler. Acta Physiologica Scandinavica 1993, 147(2): 195–201.
Biomarkers Definitions Working Group . Biomarkers and surrogate endpoints: Preferred
definitions and conceptual framework. Clinical Pharmacol Ther 2001, 69: 89–95.
Clark JA Effects of surface emissivity and viewing angle errors in thermography. Acta
Thermographica 1976, 1: 138–141.
Dzien C , Halder W , Winner H , Lechleitner M Covid-19 screening: Are forehead temperature
measurements during cold outdoor temperatures really helpful?. Wiener Klinische
Wochenschrift 2021, 133(7): 331–335.
Engel J-M , Flesch U , Stüttgen G eds. Thermological Methods. Weinheim: VCH, 1985.
Fernández-Cuevas I , Marins JCB , Lastras JA , Carmona PMG , Cano SP , García-
Concepción MÁ , Sillero-Quintana M Classification of factors influencing the use of infrared
thermography in humans: A review. Infrared Physics & Technology 2015, 71: 28–55.
Filippini C , Cardone D , Perpetuini D , Chiarelli AM , Gualdi G , Amerio P , Merla A
Convolutional neural networks for differential diagnosis of Raynaud’s phenomenon based on
hands thermal patterns. Applied Sciences 2021, 11(8): 3614.
Fisher G , Foster EB , Pascoe DD Equilibration period following exposure to hot or cold
conditions when using infrared thermography. Thermology International 2008, 18(3): 95–100.
Getson P , Schwartz RG , Brioschi M , Pittman J , Rind B , Crawford J , Usuki H , Amalu W ,
Head J Guidelines for breast thermography. Pan American Journal of Medical Thermology
2015, 2(1): 26–34.
Goodman PH , Murphy MG , Siltanen GL , Kelley MP , Rucker L Normal temperature
asymmetry of the back and extremities by computer-assisted infrared imaging. Thermology
1988, 1: 194–202.
Herrick AL , Murray A The role of capillaroscopy and thermography in the assessment and
management of Raynaud’s phenomenon. Autoimmunity Reviews 2018, 17(5): 465–472
Imeri L Thermoregulation as a non-unified system: A difficult to teach concept. Temperature
2017, 4(1): 1–8. doi: 10.1080/23328940.2017.1281872
Khallaf A. , Williams RW , Ring EFJ , Elvins DM Thermographic study of heat loss from the
face. Thermologie Österreich 1994, 4(2): 49–54.
Kish MA Guide to development of practice guidelines. Clinical Infectious Diseases 2001,
32(6): 851–854.
Kolacz S , Moderhak M , Jankau J New perspective on the in vivo use of cold stress dynamic
thermography in integumental reconstruction with the use of skin-muscle flaps. Journal of
Surgical Research 2017, 212 : 68–76.
Love TJ Heat transfer considerations in the design of a thermology clinic. Thermology 1985,
1: 88–91.
Madhu H , Kakileti ST , Venkataramani K , Jabbireddy S. Extraction of medically interpretable
features for classification of malignancy in breast thermography. In 38th Annual International
Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), IEEE 2016,
pp. 1062–1065.
Marins JCB , Moreira DG , Cano SP , Sillero Q , Soares DD , Fernandes ADA , de Silva CS ,
Costa CMA , de Amorim PR , Time required to stabilize thermographic images at rest.
Infrared Physics& Technology 2014, 65: 30–35.
Marjanovic EJ , Britton J , Howell KJ , Murray AK Quality assurance for a multi-centre
thermography study. Thermology International 2018, 28(1): 7–13.
Melnizky P , Ammer K , Rathkolb O Thermographic findings of the lower extremity in Patients
with Type II diabetes. Thermology International 2002, 12(3): 107–114.
Midttun M , Sejrsen P Blood flow rate in arterio-venous anastomoses and capillaries in
thumb, first toe, ear lobe, and nose. Clinical Physiology 1996, 44(16): 275–289.
Moreira DG , Costello JT , Brito CJ , Adamczyk JG , Ammer K , Bach AJE , Costa CMA ,
Eglin C , Fernandes AA , Fernández-Cuevas I , Ferreira JJA , Formenti D , Fournet D ,
Havenith G , Howell K , Jung A , Kenny GP , Kolosovas-Machuca ES , Maley MJ , Merla A ,
Pascoe D , Priego-Quesada JI , Schwartz RG , Seixas ARD , Selfe J , Vainer BG , Sillero-
Quintan M Thermographic imaging in sports and exercise medicine: A Delphi study and
consensus statement on the measurement of human skin temperature. Journal of Thermal
Biology 2017, 69: 155–162.
Nugent PW , Shaw JA , Pust NJ Correcting for focal-plane-array temperature dependence in
microbolometer infrared cameras lacking thermal stabilization. Optical Engineering 2013, 52
(6): 061304.
Oliver B , Munro AG , Herrington LC The effect of distance and angle of a smartphone-
compatible infrared thermal imaging camera on skin temperature at the midportion of the
Achilles tendon. Thermology International 2020, 30(2): 51–57.
Pascoe DD , Fisher G Comparison of measuring sites for the assessment of body
temperature. Thermology International 2009, 19(2): 35–42; Pascoe DD Comparison of
measuring sites for the assessment of body temperature: Final amendment. Thermology
International 2010, 20(1): 36–38.
Pauling JD , Flower V , Shipley JA , Harris ND , McHugh NJ Influence of the cold challenge
on the discriminatory capacity of the digital distal–dorsal difference in the thermographic
assessment of Raynaud’s phenomenon. Microvascular Research 2011, 82(3): 364–368.
Plassmann P , Jones CD Recording thermal images and software development for
standardisation of thermal imaging. In Wiecek B (ed.) Thermography and Lasers in Medicine.
Lodz: Akademickie Centrum Grafizno-Marketingowe Lodart S.A, 2003, pp. 19–24.
Plassmann P , Kluwe B Plantar foot assessment using liquid crystal thermography.
Thermology International 2021, 31(3): 99–100.
Plassmann P , Ring EFJ , Jones CD Quality assurance of thermal imaging systems in
medicine. Thermology International 2006, 16: 10–15.
Ring EFJ A thermographic index for the assessment of ischemia. Acta Thermographica 1980,
5: 35–38.
Ring EFJ Quality control in lnfrared thermography. In: Ring E.F.J. and Phillips B. , editors.
Recent Advances in Medical Thermology. New York: Plenum Press, 1984, pp. 185–194.
Ring EFJ , Aarts N , Black CM , Boesiger P et al. Raynaud’s phenomenon: Assessment by
thermography. Thermology 1988, 3: 69–73.
Ring EFJ , Ammer K The technique of infrared imaging in medicine. Thermology International
2000, 10(1): 7–14.
Romanovsky AA Thermoregulation: some concepts have changed. Functional architecture of
the thermoregulatory system. The American Journal of Physiology-Regulatory, Integrative
and Comparative Physiology 2007, 292: R37–R46, First published September 28, 2006;
doi:10.1152/ajpregu.00668.2006
Rost A Lehrbuch der Regulationsthermographie. Stuttgart: Hippokrates Verlag, 1993.
Schwartz RG , Elliott R , Goldberg GS , Govindan S , Conwell T , Hoekstra PP Practice
Guidelines Committee of the American Academy of Thermology: Guidelines for
neuromusculoskeletal thermography. Thermology International 2006, 16: 5–9.
Schwartz RG , Getson P , O’Young B , Campbel J , Brioschi M , Usuki H , Crawford J , Steed
P , Ammer K , Serbu G Guidelines for dental-oral and systemic health infrared thermography.
Pan American Journal of Medical Thermography 2015, 2(1): 44–53.
Schwartz RG , Horner C , Kane R , Getson P , Brioschi M , Pittman J , Rind B , Campbell J ,
Ehle E , Mustovoy A , Crawford J , Tokman A Guidelines for Breast Thermology n.d.
[Link]
accessed 19.07.2021
Simpson R , Machin G , McEvoy H , Rusby R Traceability and calibration in temperature
measurement: A clinical necessity. Journal of Medical Engineering & Technology 2006, 30:
212–217.
Stratford PW , Riddle DL Assessing sensitivity to change: Choosing the appropriate change
coefficient. Health and Quality of Life Outcomes 2005, 3(1): 23.
Turner TA , Marcella K , Riegel R , Godbold J , Muniz Honório K , Schwartz RG , Campbell J
, Brioschi M Veterinary guidelines for infrared thermography. American Academy of
Thermology, 2019. [Link]
for-infrared-thermography/ accessed 19.07.2021
Usamentiaga R , Garcia DF , Ibarra-Castanedo C , Maldague X. Highly accurate geometric
calibration for infrared cameras using inexpensive calibration targets. Measurement 2017,
112: 105–116.
Vardasca R The influence of angles and distance on assessing inner canthi of the eye skin
temperature. Thermology International 2017, 27(4): 130–135.
Wallace JD , Cade CM Clinical Thermography. Cleveland: CRC Press, 1975.
Watmough DJ , Fowler PW , Oliver R The thermal scanning of a curved isothermal surface.
Physics in Medicine & Biology 1970, 15: 1–8.
Westermann S , Stanek C , Schramel JP , Ion A , Buchner HHF. The effect of airflow on
thermographically determined temperature of the distal forelimb of the horse. Equine
Veterinary Journal 2013, 45(5): 637–641.
Weum S , Mercer JB , de Weerd L. Evaluation of dynamic infrared thermography as an
alternative to CT angiography for perforator mapping in breast reconstruction: A clinical
study. BMC Medical Imaging 2016, 16: 43.
Wiecek B , Jung A , Zuber J. Emissivity - Bottleneck and challenge for thermography.
Thermology International 2000, 10(1): 15–19.
Wilkinson JD , Leggett SA , Marjanovic EJ , Moore TL , Allen J , Anderson ME et al. A
multicenter study of the validity and reliability of responses to hand cold challenge as
measured by laser speckle contrast imaging and thermography: Outcome measures for
systemic sclerosis–related Raynaud’s phenomenon. Arthritis & Rheumatology 2018, 70(6):
903–991.
Woodrough RE Medical Infrared-Thermography. Principles and Practice. New York:
Cambridge University Press, 1982.
Zhang HY , Youk T , Lee HK , Song HJ , Yang KH Reference standard temperature data of
normal Korean extremities. Thermology International 2019, 29(2): 67–72.

Basic Approaches of Artificial Intelligence and Machine Learning in

Thermal Image Processing
Allaberdiev S , Yakhyoev S , Fatkhullayev R , Chen J Speeded-up robust feature matching
algorithm based on image improvement technology. Journal of Computer and
Communications 2019, 7, 1–10. [Link]
Batchuluun G , Yoon HS , NGuyen DT , Pham TD , Park KR . A study on the elimination of
thermal reflections. IEEE Access 2019, 7, 174597–174611.
[Link]
Bay H , Tuytelaars T , and Van Gool L. 2006. “SURF: Speeded Up Robust Features.” In
Computer Vision – ECCV 2006, 404–417. Berlin, Heidelberg: Springer.
Erickson BJ , Korfiatis P , Akkus Z , Kline TL . Machine learning for medical imaging.
Radiographics 2017, 37(2): 505–515.
Erwin E. “Peer Review Performance Analysis of Comparison between Region Growing,
Adaptive Threshold and Watershed Methods for Image Segmentation.” (2020).
Ester M , Kriegel H-P , Sander J , Xu X A density-based algorithm for discovering clusters in
large spatial databases with noise. In E. Simoudis , J. Han , & U.M. Fayyad (Eds.),
Proceedings of the second International Conference on Knowledge Discovery and Data
Mining. Portland: AAAI Press. 1996.
Friedman J , Hastie T , Tibshirani R Additive logistic regression: A statistical view of boosting.
The Annals of Statistics 28(2): (2000), 337–374.
Fusco R , Piccirillo A , Sansone M , Granata V , Rubulotta MR , Petrosino T , Barretta ML ,
Vallone P , Di Giacomo R , Esposito E , Di Bonito M , Petrillo A. Radiomics and artificial
intelligence analysis with textural metrics extracted by contrast-enhanced mammography in
the breast lesions classification. Diagnostics 2021, 11(5): 815.
[Link]
Gao H , Liu Z , Van Der Maaten L , Weinberger KQ . Densely Connected Convolutional
Networks. In: Proceedings - 30th IEEE Conference on Computer Vision and Pattern
Recognition, CVPR 2017. [Link]
Ghosh A , Sufian A , Sultana F , Chakrabarti A , De D Fundamental concepts of convolutional
neural network. In: Recent Trends and Advances in Artificial Intelligence and Internet of
Things. Springer, Cham, 2020. 519–567. [Link]
Gonzalez RC , Woods RE Digital Image Processing, 4th Edition, Pearson. 2018, Accessed
August 27, 2021 . [Link]
Image-Processing-4th-Edition/[Link]
Goodfellow I , Bengio Y , Courville A Deep Learning. The MIT Press 2016. Accessed August
27, 2021 . [Link]
Guo G , Wang H , Yan Y , Zheng J , Li B A fast face detection method via convolutional
neural network. Neurocomputing 2015, 395(Bo Li): 128–137.
[Link]
Guo Y , Sengur A NCM: Neutrosphic c-means clustering algorithm. Pattern Recognition
2015, 48: 2710–2724.
Harris C , Stephens M A combined corner and edge detector, in Alvey Vision Conference,
vol. 15, Manchester, UK, 1988, pp. 10–5244.
He K , Zhang X , Ren S , Sun J Deep Residual Learning for Image Recognition. Proceedings
of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
2016-December, 2015 (December): 770–778. [Link]
Hemalatha RJ , Thamizhvani TR , Josephin Arockia Dhivya A , Joseph JE , Babu B ,
Chandrasekran R Active contour based segmentation techniques for Medical image analysis.
In: Robert Koprowski (Eds), Medical and Biological Image Analysis, 2018,
[Link]
Howard AG , Zhu M , Bo C , Kalenichenko D , Wang W , Weyand T , Andreetto M , Adam H
MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications. April.
2017. [Link]
Huang G , Liu S , Van der Maaten L , Weinberger KQ Condensenet: An efficient densenet
using learned group convolutions. In Proceedings of the IEEE conference on computer vision
and pattern recognition, pp. 2752–2761. 2018.
Hubel DH , Wiesel TN Receptive fields and functional architecture of monkey striate cortex.
The Journal of Physiology 1968, 195 (1): 215–243.
[Link]
John V , Tsuchizawa S , Liu Z , Mita S Fusion of thermal and visible cameras for the
application of pedestrian detection. Signal, Image and Video Processing 2017, 11(3):
517–524. [Link]
Kabir H , Jabid T , Chae O. Local directional pattern variance (LDPv): A robust feature
descriptor for facial expression recognition: The international arab Journal of Information
Technology 2012, 9(4): 382–391.
Krizhevsky A , Sutskever I , Hinton GH ImageNet classification with deep convolutional
neural networks. Communications of the ACM 2017, 60(6): 84–90.
Lowe DG Distinctive image features from scale-invariant keypoints. International Journal of
Computer Vision 60: 91–110 (2004). [Link]
Ma J , Liang P , Yu W , Chen C , Guo X , Wu J , Junjun JJ Infrared and visible image fusion
via detail preserving adversarial learning. Information Fusion 2020, 54: 85–98.
[Link]
Ma J , Ma Y , Li C. Infrared and visible image fusion methods and applications: A survey.
Information Fusion 2019, 45: 153–178. [Link]
Mall PK , Singh PK , Yadav D GLCM based feature extraction and medical X-RAY image
classification using machine learning techniques, IEEE Conference on Information and
Communication Technology, 2019, 2019, pp. 1–6, doi: 10.1109/CICT48419.2019.9066263
Mittal H , Pandey AC , Saraswat M et al. A comprehensive survey of image segmentation:
Clustering methods, performance parameters, and benchmark datasets. Multimed Tools
Application 2021. [Link]
Muthukrishnan R , Radha M. Edge detection techniques for image segmentation.
International Journal of Computer Science and Information Technology, 2011, 3.
[Link]
Nanni L , Lumini A , Brahnam S Local binary patterns variants as texture descriptors for
medical image analysis, Artificial Intelligence in Medicine, 2010, 49: 117–125.
Öztürk S , Akdemir B Application of feature extraction and classification methods for
histopathological image using GLCM, LBP, LBGLCM, GLRLM and SFTA. Procedia Computer
Science 2018, 132: 40–46.
Pajares G , de la Cruz JM A wavelet-based image fusion tutorial. Pattern Recognition 2004,
37(9): 1855–1872. [Link]
Panetta K , Wan Q , Agaian S , Rajeev S , Kamath S , Rajendran R , Wan Q , Agaian S , Rao
SP , Kaszowska A , Taylor HA , Samani A , Yuan X , et al. A comprehensive database for
benchmarking imaging systems. IEEE Transactions on Pattern Analysis and Machine
Intelligence 2020, 42(3): 509–520. [Link]
Patil A , Rane M Convolutional neural networks: An overview and its applications in pattern
recognition. Smart Innovation, Systems and Technologies 2021, 195: 21–30.
[Link]
Pattanayak S Pro Deep Learning with TensorFlow. Packt Publishing Ltd, Apress, Berkeley,
CA, 2017. [Link]
Rajinikanth V , Raj ANJ , Thanaraj KP , Naik GR . A customized VGG19 network with
concatenation of deep and handcrafted features for brain tumor detection. Applied Sciences
2020, (10): 3429. [Link]
Ronneberger O , Fischer P , Brox T U-Net: Convolutional networks for biomedical image
segmentation, 2015, [Link]
Russakovsky O , Deng J , Su H , Krause J , Satheesh S , Ma S , Huang Z , et al. ImageNet
large scale visual recognition challenge. International Journal of Computer Vision 2015,
115(3): 211–252. [Link]
Sander J , Ester M , Kriegel H-P , Xu X Density-based clustering in spatial databases: The
algorithm GDBSCAN and its applications. Data Mining and Knowledge Discovery 1998, 2(2):
169–194.
Saparudin E , Nevriyanto A , Purnamasari D Performance analysis of comparison between
region growing, adaptive threshold and watershed methods for image segmentation.
Proceedings of the International MultiConference of Engineers and Computer Scientists
2018, 2018, Hong Kong.
Sappa AD , Carvajal JA , Aguilera CA , Oliveira M , Romero D , Vintimilla B. Wavelet-based
visible and infrared image fusion: A comparative study. Sensors 2016, 16: 861.
[Link]
Sargent D , Chen C-I , Tsai C-M , Wang Y-F , Koppel D. Feature detector and descriptor for
medical images. Proceedings SPIE 7259 Medical Imaging 2009: Image Processing, 2009.
[Link]
Sharma N , Aggarwal LM Automated medical image segmentation techniques. Journal of
Medicine and Physics 2010, 35: 3–14. [Link]
Simonyan K , Zisserman A. Very deep convolutional networks for large-scale image
recognition 2014 arXiv preprint arXiv:1409.1556,2014
Teledyne FLIR. n.d. “FLIR E54 Advanced Thermal Imaging Camera | Teledyne FLIR.”
Accessed August 27, 2021 . [Link]
Vollmer M , Möllmann KP . Infrared Thermal Imaging: Fundamentals, Research and
Applications, 2nd Edition, n.d., John Wiley & Sons, Germany.
Wernick MN , Yang Y , Brankov JG , Yourganov G , Strother SC Machine learning in Medical
imaging. IEEE Signal Processing Magazine 2010, 27(4): 25–38.
Wu H , Huang J , Wang X , Jia J Image Registration of Infrared and Visible Based on SIFT
and SURF. In: Tenth International Conference on Digital Image Processing (ICDIP 2018),
edited by Jiang X and Hwang J.N. , 2018, 10806:186. SPIE.
[Link]

Thermal Imaging for Arthritis Evaluation in a Small Animal Model

Almarestani L , Fitzcharles MA , Bennett GJ , da-Silva AR . Imaging studies in Freund’s
complete adjuvant model of regional polyarthritis, a model suitable for the study of pain
mechanisms, in the rat. Arthritis & Rheumatism 2011; 63(6): 1573–1581.
Banerjee S , Webber C , Poole AR . The induction of arthritis in mice by the cartilage
proteoglycan aggrecan: Roles of CD4+ and CD8+ T cells. Cell Immunol 1992; 144:347–357.
Brenner M , Braun C , Oster M , Gulko PS . Thermal signature analysis as a novel method for
evaluating of inflammatory arthritis activity. Ann Rheum Dis 2006; 65(3):306–311.
Brenner M , Meng HC , Yarlett NC , Griffiths MM , Remmers EF , Wilder RL , Gulko PS The
non-MHC quantitative trait locus Cia 10 contains a major arthritis gene and regulates disease
severity, pannus formation and joint damage. Arthritis Rheumatism 2005; 52(1): 322–332.
Bresnihan B Are synovial biopsies of diagnostic value?. Arthritis Research and Therapy
2003; 5:271–278.
Calkosinski I , DobrzyNski M RosiNczuk J , Dudek K , Chrószcz A , Fita K , Dymarek R. The
use of infrared thermography as a rapid, quantitative, and non-invasive method for evaluation
of inflammation response in different anatomical regions of rats. BioMed Research
International. 2015, Article ID 972535.
Chan, Gray BD , Pak KY , Fong D. Non-invasive in vivo imaging of arthritis in a collagen-
induced murine model with phosphatidylserine-binding near-infrared (NIR) dye. Arthritis
Research & Therapy 2015; 17:50.
Choi Y , Yoon YW , Na HS , Kim SH , Chung JM . Behavioural signs of ongoing pain and cold
allodynia in a rat model of neuropathic pain. Pain 1994; 59:369–376.
Dai Q , Zhou D , Xu L , Song X. Curcumin alleviates rheumatoid arthritis-induced
inflammation and synovial hyperplasia by targeting mTOR pathway in rats. Drug Des Devel
Ther 2018; 12:4095–4105.
Hargreaves K , Dubner R , Brown F , Flores C , Joris J. A new and sensitive method for
measuring thermal nociception in cutaneous hyperalgesia. Pain 1988; 32:77–88:32.
Jasemian Y , Svendsen P , Deleuran B , Dagnees-Hansen F. Refinement of the collagen
induced arthritis model in rats by infrared thermography. British journal of medicine and
Medical Research 2011; 1(4):469–477.
Kim HY , Kim WU , Cho ML , Lee SK , Youn J , Kim SI , et al. Enhanced T cell proliferative
response to type II collagen and synthetic peptide CII (255-274) in patients with rheumatoid
arthritis. Arthritis Rheum 1999; 42:2085–2093.
Kim WU , Cho ML , Jung YO , Min SY , Park SW , Min DJ , et al. Type II collagen
autoimmunity in rheumatoid arthritis. Am J Med Sci 2004; 327:202–211.
Koizumi F , Matsuno H , Wakakaki K Synovitis in rheumatoid arhtrtis: scoring of characteristic
histopathological features. Pathology International 1999; 49:298–304.
Krenn V , Morawietz L , Burmester GR , Kinne RW , Mueller-Ladner U , Muller B , Haupl T.
Synovitis score: discrimination between chronic low-grade and high-grade synovitis.
Histopathology 2006; 49:358–364.
Lim MA , Louie B , Ford D , Heath K , Cha P , Lacroix JB , Lum PY , Robertson TL ,
Schaevitz L. Development of disease arthritis index, a novel metric to measure disease
parameters in a Rat model of Rheumatoid arthritis. Front Pharmocol 2017; 8:818.
Mikecz K , Glant TT , Buzás E , Poole AR . Proteoglycan-induced polyarthritis and spondylitis
adoptively transferred to naive (nonimmunized) BALB/c mice. Arthritis Rheum 1990;
33:866–876.
Nosratiz, BM , Rodriguez CR , Saatchii K , Hafeli UO Refinement and Validation of infrared
thermal imaging (IRT): A non-invasive technique to measure disease activity in a mouse
model of rheumatoid arthritis. Arthritis Research & Therapy 2020; 22:281.
Pearson CM . Development of arthritis, periarthritis and periostitis in rats given adjuvants.
Proceedings of the Society for Experimental Biology and Medicine 1956; 91(1):95–101.
Sanchez BM , Lesch M , Brammer D , Bove SE , Thiel M , Kilgore KS . Use of a portable
thermal imaging unit as a rapid, quantitative method of evaluating inflammation and
experimental arthritis. J Pharmacol Toxicol Methods 2008; 57(3):169–175.
Snekhalatha U , Anburajan M , Venkatraman B , Menaga M , Raj B. Evaluation of Complete
Freund Adjuvant Induced Arthrtitis model in Wistar rats by Thermography, A comparison with
histopathology. Zeitschrift fur Rheumatologie 2012; 72(4):375–382.
Tal M , Bennett GJ . Extra-territorial pain in rats with a peripheral mononeuropathy: mechano-
hyperalgesia and mechano-allodynia in the territory of an uninjured nerve. Pain 1994;
57:375–382.
Tuncel J , Haag S , Hoffmann MH , Yau ACY , Hultqvist M , Olofsson P , et al. Animal models
of rheumatoid arthritis (I): Pristane-induced arthritis in the rat. PLoS ONE 2016;
11(5):e0155936. doi:10.1371/[Link].0155936
Wooley PH , Siegner SW , Whalen JD , Karvonen RL , Fernández-Madrid F. Dependence of
proteoglycan induced arthritis in BALB/c mice on the development of autoantibodies to high
density proteoglycans. Ann Rheum Dis 1992; 51:983–991.
Xia C , Li J , Yue Y , Chen L , Dai J. Biophoton imaging evaluation of the process of
rheumatoid arthritis in rats. Natural Science, 2021; 13(10):451–456.
Yu Y , Xiong Z , Lv Y , Qian Y , Jiang S , Tian Y. In vivo evaluation of early disease
progression by X-ray phase-contrast imaging in the adjuvant-induced arthritic rat Skeletal
Radiol 2006; 35:156–164.

Thermal Imaging for Inflammatory Arthritis Evaluation

Aletaha D , Neogi T , Silman AJ , Funovits J , Felson DT , Bingham III CO et al. 2010
rheumatoid arthritis classification criteria: An American College of Rheumatology/European
League Against Rheumatism collaborative initiative. Arthritis & Rheumatology 2010, 62(9):
2569–2581.
Ammer K and Ring F. The Thermal Human Body. A Practical Guide to Thermal Imaging.
Jenny Standord Publishing, Singapore, 2019.
Anderson JK , Zimmerman L , Caplan L , Michaud K Measures of rheumatoid arthritis
disease activity. Arthritis Care and Research 2011, 63(S119): S14–S36.
Arnett FC , Edworthy SM , Bloch DA , McShane DJ , Fries JF , Cooper NS , et al. The
American Rheumatism Association 1987 revised criteria for the classification of rheumatoid
arthritis. Arthritis & Rheumatology 1988, 31: 315–324.
Beitinger N , Ehrenstein B , Schreiner B , Fleck M , Grifka J , Lüring C , Hartung W et al. The
value of colour doppler sonography of the knee joint: A useful tool to discriminate
inflammatory from non-inflammatory disease? Rheumatology 2013, 52(8): 1425–1428.
Chaudhary AS , Chaturvedi DK Efficient thermal image segmentation for heat visualization in
solar panels and batteries using watershed transform I.J. International Journal of Image,
Graphics and Signal Processing 2017, 11: 10–17.
Chen W , Cheng H , Shen H An effective methodology for thermal characterization of
electronic packaging. IEEE Transactions on Components, Packaging and Manufacturing
Technology 2003, 26: 222–232.
Colebatch AN , et al. EULAR recommendations for the use of imaging of the joints in the
clinical management of rheumatoid arthritis. Annals of the Rheumatic Diseases 2012, 72(6):
804–814. doi:10.1136/annrheumdis-2012-203158
Collins AJ , Ring EFJ , Cosh JA , Bacon PA Quantitation of thermography in arthritis using
multi-isothermal analysis. Annals of the Rheumatic Diseases Disorders 1974, 33: 113–115.
Denoble AE , Hall N , Pieper CF , Kraus VB Patellar skin surface temperature by
thermography reflects knee osteoarthritis severity. Clinical Medicine Insights: Arthritis and
Musculoskeletal Disorders 2010, 3: 69–75.
Devereaux MD , Parr GR , Lachmann SM , Page Thomas DP , Hazleman BL Thermographic
diagnosis in athletes with patellofemoral Arthralgia. Journal of Bone and Joint Surgery 1986,
68(1): 42–44.
Devereaux MD , Parr GR , Page Thomas DP , Hazleman BL Disease activity indexes in
rheumatoid arthritis; a prospective comparative study with thermography. Annals of the
Rheumatic Diseases Disorders 1985, 44: 434–437.
Duarte A , Carrão L , Espanha M , Viana T , Freitas D , Bártolo P , Faria P , Almeida HA
Segmentation algorithms for thermal images. Procedia Technology 2014, 16: 1560–1569.
Fiocco U , Cozzi L , Rubaltelli L , et al. Long-term sonographic follow-up of rheumatoid and
psoriatic proliferative knee joint synovitis. British Society for Rheumatology 1996, 35:
155–163.
Fries J , Spitz PW , Young DY The dimensions of health outcomes: The health assessment
questionnaire, disability and pain scales. The Journal of Rheumatology 1982, 9: 789–793.
Frize M , Adea C , Payeur P , Di Primio G , Karsh J , Ogungbemile A Detection of rheumatoid
arthritis using infrared imaging. Proceedings of SPIE 2011, 7962.
Ghosh A , Ghosh B , Pain S , Pande A , Saha S , Banarjee A , Biswas A Comparison
between DAS28, CDAI and HAQ-DI as tools to monitor early rheumatoid arthritis patients in
eastern India. International Journal of Rheumatology 2011, 6(3): 116–122.
Gonzalez RC , Woods RE Digital Image Processing 3’d Edition, Prentice Hall, 2007,
598–603.
Heidari B Rheumatoid arthritis: Early diagnosis and treatment outcomes. Caspian Journal of
Internal Medicine winter; 2011, 2(1): 161–170.
Wayne Rasband , Image Processing and Analysis in Java (ImageJ) Software Documentation,
1997, [Link]
Jadin M Recent progress in diagnosing the reliability of electrical equipment. Infrared Physics
and Technology 2012, 55(4): 236–245.
Kacmaz S , Ercelebi E , Zengin S , Cindoruk S The use of infrared thermal imaging in the
diagnosis of deep vein thrombosis. Infrared Physics and Technology 2017, 86: 120–129.
Kadir T , Brady M Saliency, scale and image description. International Journal of Computer
Vision 2001, 45: 83–105.
Krishnan E , Sokka T , Häkkinen A , Hubert H , Hannonen P Normative values for the health
assessment questionnaire disability index. Arthritis & Rheumatology 2004, 50: 953–960.
Lasanen R , et al. , Thermal imaging in screening of joint inflammation and rheumatoid
arthritis in children. Physiological Measurement 2015, 36: 273.
Likas A , Vlassis N , Verbeek JJ The global k-means clustering algorithm. Pattern
Recognition 2003, 36(2): 451–461. doi:10.1016/s0031-3203(02)00060-2
Luo C , Yang W , Huang P , Zhou J Overview of image matching based on ORB algorithm.
IOP Conference Series 2019, 1237: 032020.
Maeda M , Kachi H , Ichihashi N , Oyama Z , Kitajima Y The effect of electrical acupuncture
stimulation therapy using thermography and plasma endothelin levels in patients with
progressive systemic sclerosis. Journal of Dermatology Science 1998, 17(2): 151–155.
Mishra R , Sharma BL , Gupta R , Pandya S , Agarwal S , Agarwal P , Grover S , Sarma P ,
Wanjam K Indian Rheumatology Association consensus statement on the management of
adults with rheumatoid arthritis. Indian Journal of Rheumatology 2008, 3(3): S1–S16.
Mittal A , Dubey SK Analysis of MRI images of Rheumatoid Arthritis through morphological
image processing techniques. International Journal of Computing Science 2013, 10(3):
118–122.
Nameirakpam D , Khumanthem M , Yambem JC Image segmentation using K-means
clustering algorithm and subtractive clustering algorithm. 11th International Conference on
Signal, Image Processing and Pattern Recognition, 2015, 54: 764–771, August.
Nandagopan GL , Haripriya AB Implementation and comparison of two image segmentation
techniques on thermal foot images and detection of ulceration using asymmetry. 2016
International Conference on Communication and Signal Processing (ICCSP), Melmaruvathur,
India, 2016: 0356–0360.
Nandhakumar N , Aggarwal JK Integrated analysis of thermal and visual images for scene
interpretation. IEEE Transaction on Pattern Analysis and Machine Intelligence 1988, 10(4):
469–481.
Online disease activity score calculator for rheumatoid arthritis , available from
[Link] Accessed on 2021.
Preetha MMSJ , Suresh LP , Bosco MJ Image segmentation using seeded region growing.
IEEE International Conference on Computing, Electronics and Electrical Technologies
(ICCEET) 2012, 576–583.
Qvistgarrd E , Rogind H , Torp-Pedersen S , et al. Quantitative ultrasonography in
rheumatoid arthritis: Evaluation of inflammation by doppler technique. Annals of the
Rheumatic Diseases 2001, 60: 690–693.
Ramírez-Rozo TJ , García-Álvarez JC , Castellanos-Domínguez CG Infrared thermal image
segmentation using expectation-maximization-based clustering. 2012 XVII Symposium of
Image Signal Processing, and Artificial Vision (STSIVA), 2012, 223–226. doi:
10.1109/STSIVA.2012.6340586
Ring EFJ Quantitative thermography and thermographic index. Verh Dtsch Ges
Rheumatology 1980, 6: 287–288.
Roemer FW , Zhang Y , Niu J et al. Tibiofemoral joint osteoarthritis: Risk factors for MR-
depicted fast cartilage loss over a 30-month period in the multicenter osteoarthritis study.
Radiology 2009, 252: 772–780.
Salisbury R , Parr G , De Silva M , Hazleman BL , Page-Thomas DP Heat distribution over
normal and abnormal joints: thermal pattern and quantification. Annals of the Rheumatic
Diseases 1983, 42(5): 494–499.
Schmidt WA , Völker L , Zacher J , Schläfke M , Ruhnke M , Gromnica-Ihle E Colour doppler
ultrasonography to detect pannus in knee joint synovitis. Clinical and Experimental
Rheumatology 2000, 2000(18): 439–444.
Sergyan S Edge detection techniques of thermal images IEEE 10th Jubilee International
Symposium on Intelligent Systems and Informatics Subotica, Serbia 2012.
Shaikh S , Gite H , Manza RR , Kale KV , Akhter N Segmentation of thermal images using
thresholding-based methods for detection of malignant tumours. Intelligent Systems:
Technology and Applications 2016, 2016: 131–146.
Snekhalatha U , Anburajan M , Venkatraman B , Menaka M , Raj B Evaluation of rheumatoid
arthritis in small animal model using thermal imaging. Proceedings of the International
Conference Signal Processing Communication Computing and Networking Technologies
2011, 785–791.
Snekhalatha U , Anburajan M , Sowmiya V , Venkatraman B , Menaka M Automated hand
thermal image segmentation and feature extraction in the evaluation of rheumatoid arthritis.
Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in
Medicine 2015, 229(4): 319–331.
Snekhalatha U , Sowmiya V , Nilkantha G A computer aided diagnostic based Thermal image
Analysis: A potential tool for the Evaluation of Rheumatoid Arthritis in Hand. Journal of
Medical and Biological Engineering 2017, 38(4): 666–677.
Spalding SJ , Kwoh CK , Boudreau R , Enama J , Lunich J , Huber D , Denes L , Hirsch R
Three dimensional and thermal surface imaging produces reliable measures of joint shape
and temperature: A potential tool for quantifying arthritis. Arthritis Research & Therapy 2008,
10(1): R10. doi:10.1186/ar2360
Suma AB , Snekhalatha U , Rajalakshmi T Automated thermal image segmentation of knee
rheumatoid arthritis. 2016 International Conference on Communication and Signal
Processing (ICCSP) (2016), pp. 0535–0539.
Wang W , Zhang J , Shen C Improved human detection and classification in thermal images.
IEEE International Conference on Image Processing 2010, 2313–2316.
Warashina H , Hasegawa Y , Tsuchiya H , Kitamura S , Yamauchi KI , Torii Y , Kawaski M ,
Sakano S Clinical radiographic and thermographic assessment of osteoarthritis in the knee
joints. Annals of the Rheumatic Diseases 2002, 61(9): 852–854.

Potential of Thermal Imaging to Detect Complications in Diabetes

Abdelalim EM Modeling different types of diabetes using human pluripotent stem cells.
Cellular and Molecular Life Sciences 2020 78:6 2020, 78(6): 2459–2483.
[Link]
Adam M , Ng EYK , Shu Lih O , Heng ML , Hagiwara Y , Tan JH , Tong JWK , Rajendra
Acharya U Automated characterization of diabetic foot using nonlinear features extracted
from thermograms. Infrared Physics and Technology 2018, 89: 325–337.
[Link]
Alzubaidi L , Fadhel MA , Al-Shamma O , Zhang J , Santamaría J , Duan Y , Oleiwi SR
Towards a better understanding of transfer learning for medical imaging: A case study.
Applied Sciences 2020, 10(13): 4523. [Link]
Ammer K The Glamorgan protocol for recording and evaluation of thermal images of the
human body. Thermology International 2008, 18(4): 125–144.
Ammer K , Melnizky P , Rathkolb O , Ring EF . Thermal Imaging of Skin Changes on the
Feet of Type II Diabetics. In 2001 Conference Proceedings of the 23rd Annual International
Conference of the IEEE Engineering in Medicine and Biology Society, 2001, 3:2870–2872.
IEEE. [Link]
Ammer K , Ring F The Thermal Human Body. The Thermal Human Body. Jenny Stanford
Publishing, 2019. [Link]
Ashwin Kumaar M , Thanaraj P Feature extraction of arterio-venous malformation images
using grey level co-occurrence matrix. Indian Journal of Science and Technology 2015,
8(35): 5. [Link]
Bay H , Ess A , Tuytelaars T , Van GL . Speeded-up robust features (SURF). Comput Vis
Image Underst 2008, 110(3): 346–359. [Link]
Brånemark PI , Fagerberg SE , Langer L , Säve-Söderbergh J Infrared thermography in
diabetes mellitus a preliminary study. Diabetologia 1967, 3(6): 529–532.
[Link]
Lowe D Distinctive image features from scale-invariant keypoints. International Journal of
Computer Vision, 2004a, 60(2): 91–110.
Cruz-Vega I , Hernandez-Contreras D , Peregrina-Barreto H , de Jesus Rangel-Magdaleno J
, Ramirez-Cortes JM Deep learning classification for diabetic foot thermograms. Sensors
2020, 20(6): 1762. [Link]
Gavin JR , Alberti KGMM , Davidson MB , Defronzo RA , Drash A , Gabbe SG , Genuth S ,
Harris MI , Kahn R , Keen H , Knowler WC , Lebovitz H , Maclaren NK , Palmer JB , Raskin P
, Rizza RA , Stern MP . Expert committee on the diagnosis and classification of diabetes
mellitus: Report of the expert committee on the diagnosis and classification of diabetes
mellitus. Diabetes Care 2003, 26(Suppl. 1): S5–S20.
Facchinetti A , Del Favero S , Sparacino G , Castle JR , Kenneth Ward W , Cobelli C
Modeling the glucose sensor error. IEEE Transactions on Biomedical Engineering 2014,
61(3): 620–629. [Link]
Francis Ring EJ History of thermology and thermography: Pioneers and progress.
Thermology International 2012, 22(3).
Harding JR , Wertheim DF , Williams RJ , Melhuish JM , Banerjee D , Harding KG Infrared
imaging in diabetic foot ulceration. 2002, November, 916–918.
[Link]
Hernandez-Contreras D , Peregrina-Barreto H , Rangel-Magdaleno J , Gonzalez-Bernal JA ,
Altamirano-Robles L A quantitative index for classification of plantar thermal changes in the
diabetic foot. Infrared Physics & Technology 2017, 81(March): 242–249.
[Link]
Hernandez-Contreras DA , Peregrina-Barreto H , Rangel-Magdaleno De Jesus J , Renero-
Carrillo FJ Plantar thermogram database for the study of diabetic foot complications. IEEE
Access 2019, 7: 161296–161307. [Link]
Ilo A , Romsi P , Mäkelä J Infrared thermography and vascular disorders in diabetic feet.
Journal of Diabetes Science and Technology 2020, 14(1): 28–36.
[Link]
Amalu W International academy of clinical thermology quality assurance guidelines standards
and protocols in clinical thermographic imaging. Eagle Institute of Clinical Thermology 2020,
2020.
Kaabouch N , Chen Y , Anderson J , Ames F , Paulson R Asymmetry analysis based on
genetic algorithms for the prediction of foot ulcers. Visualization and Data Analysis 2009
2009, 7243 (January): 724304. [Link]
Karami E , Prasad S , Shehata M Image matching using SIFT, SURF, BRIEF and ORB:
Performance comparison for distorted images. 2017, October.
[Link]
Lavery LA , Armstrong DG Temperature monitoring to assess, predict, and prevent diabetic
foot complications. Current Diabetes Reports 2007, 7(6): 416–419.
[Link]
Lavery LA , Higgins KR , Lanctot DR , Constantinides GP , Zamorano RG , Armstrong DG ,
Athanasiou KA , Mauli Agrawal C Home monitoring of foot skin temperatures to prevent
ulceration. Diabetes Care 2004, 27(11): 2642–2647.
[Link]
Liu C , van Netten JJ , van Baal JG , Bus SA , van der Heijden F Automatic detection of
diabetic foot complications with infrared thermography by asymmetric analysis. Journal of
Biomedical Optics 2015, 20(2): 026003. [Link]
Lowe DG Distinctive image features from scale-invariant keypoints. International Journal of
Computer Vision 2004 60:2 2004b, 60(2): 91–110.
[Link]
Raghu M , Zhang C , Kleinberg J , Bengio S. Transfusion: Understanding transfer learning
with applications to medical imaging. 2019. [Link]
Renero CFJ The thermoregulation of healthy individuals, overweight–obese, and diabetic
from the plantar skin thermogram: a clue to predict the diabetic foot. Diabetic Foot and Ankle
2017, 8(1). [Link]
Ring EFJ , Ammer K Infrared thermal imaging in medicine. Physiological Measurement 2012,
33(3): R33. [Link]
Rother KI . Diabetes treatment — Bridging the divide. The New England Journal of Medicine
2007, 356(15): 1499-1501. [Link]
Rublee E , Rabaud V , Konolige K , Bradski G ORB: An efficient alternative to SIFT or SURF.
Proceedings of the IEEE International Conference on Computer Vision 2011, 2564–2571.
[Link]
Tareen SAK , Saleem Z A Comparative Analysis of SIFT, SURF, KAZE, AKAZE, ORB, and
BRISK. 2018 International Conference on Computing, Mathematics and Engineering
Technologies: Invent, Innovate and Integrate for Socioeconomic Development, ICoMET 2018
- Proceedings 2018-January (April), 2018: 1–10.
[Link]
Shamim Hossain M , Muhammad G , Guizani N Explainable AI and mass surveillance
system-based healthcare framework to combat COVID-I9 like Pandemics. IEEE Network
2020, 34(4): 126–132. [Link]
Simonyan K , Zisserman A. VGG-16.” ArXiv Preprint. 2014.
Thirunavukkarasu U , Umapathy S , Janardhanan K , Thirunavukkarasu R. A computer aided
diagnostic method for the evaluation of Type II diabetes mellitus in facial thermograms.
Physical and Engineering Sciences in Medicine 2020 43:3 2020a, 43(3): 871–888.
[Link]
Thirunavukkarasu U , Umapathy S , Krishnan PT , Janardanan K Human tongue
thermography could be a prognostic tool for prescreening the Type II diabetes mellitus.
Evidence-Based Complementary and Alternative Medicine 2020b, 2020(January): 1–16.
[Link]
van Netten JJ , van Baal JG , Liu C , van der Heijden F , Bus SA Infrared thermal imaging for
automated detection of diabetic foot complications. Journal of Diabetes Science and
Technology 2013, 7(5): 1122–1129. [Link]
Vardasca R , Magalhaes C , Mendes J Biomedical applications of infrared thermal imaging:
current state of machine learning classification. Proceedings 2019, 27(1): 46.
[Link]
Vardasca R , Vaz L , Mendes J Classification and decision making of medical infrared
thermal images. Lecture Notes in Computational Vision and Biomechanics 2018, 26: 79–104.
[Link]
Vollmer M , Möllmann K-P Infrared thermal imaging: fundamentals, research and
applications. 2nd edition, Wiley, 794 pages, 2018.
Zimmet PZ , Magliano DJ , Herman WH , Shaw JE Diabetes: A 21st century challenge. The
Lancet Diabetes & Endocrinology 2014, 2(1): 56–64. [Link]
8587(13)70112-8

Thermal Imaging in Detection of Fever for Infectious Diseases

Ammer K , Ring F The Thermal Human Body. The Thermal Human Body. Jenny Stanford
Publishing, 2019. [Link]
Chiu WT , Lin PW , Chiou HY , Lee WS , Lee CN , Yang YY , Lee HM , et al. Infrared
thermography to mass-screen suspected SARS patients with fever. Asia-Pacific Journal of
Public Health 2005, 17(1): 26–28. [Link]
Cho KS , Yoon J Fever screening and detection of febrile arrivals at an international airport in
Korea: Association among self-reported fever, infrared thermal camera scanning, and
tympanic temperature. Epidemiology and Health 2014, 36(1): e2014004.
Chunming Wu , Zhang Y MTCNN and FACENET based access control system for face
detection and recognition. Automatic Control and Computer Sciences 2021, 55(1): 102–112.
[Link]
Déniz O , Bueno G , Salido J , De la Torre F Face recognition using histograms of oriented
gradients. Pattern Recognition Letters 2011, 32(12): 1598–1603.
[Link]
Doran G Introduction to temperature screening principles - Irish manufacturing research.
[Link] 2020. [Link]/2020/05/20/Comparison-of-Temperature-Screening-Technologies-IMR-
[Link]
Du J High-precision portrait classification based on MTCNN and its application on similarity
judgement. Journal of Physics: Conference Series 2020, 1518(April): 012066.
[Link]
El-Radhi AS Fever in common infectious diseases. In: Clinical Manual of Fever in Children,
Springer International Publishing, 2018, pp. 85–140. [Link]
92336-9_5
Ghassemi P , Joshua Pfefer T , Casamento JP , Simpson R , Wang Q Best practices for
standardized performance testing of infrared thermographs intended for fever screening.
PLoS One 2018, 13(9): e0203302. [Link]
Huang J , Shang Y , Chen H Improved viola-jones face detection algorithm based on
HoloLens. Eurasip Journal on Image and Video Processing 2019, 2019(1): 1–11.
[Link]
IUPS Thermal Commission Glossary of terms for thermal physiology. The Japanese Journal
of Physiology 2001, 51(2): 245–280.
Jiang H , Learned-Miller, E Face detection with the faster R-CNN. 2017 12th IEEE
International Conference on Automatic Face & Gesture Recognition (FG 2017), 2017,
650–657. IEEE. [Link]
Khaksari K , Nguyen T , Hill B , Quang T , Perreault J , Gorti V , Malpani R et al Review of the
Efficacy of Infrared Thermography for Screening Infectious Diseases with Applications to
COVID-19. Journal of Medical Imaging (Bellingham, Wash.) 2021, 8(Suppl 1): 010901.
[Link]
Ma M , Wang J Multi-View Face Detection and Landmark Localization Based on MTCNN.
2018 Chinese Automation Congress (CAC), 2018, 4200–4205. IEEE.
[Link]
Mackowiak PA , Wasserman SS , Levine MM A critical appraisal of 98.6°F, the upper limit of
the normal body temperature, and other legacies of carl reinhold august Wunderlich. The
Journal of the American Medical Association 1992, 268(12): 1578–1580.
[Link]
Mackowiak PA , Worden G Carl Reinhold august wunderlich and the evolution of clinical
thermometry. Clinical Infectious Diseases 1994, 18(3): 458–458.
[Link]
Marjanovic EJ . Britton J , Howell KJ , Murray AK . Quality assurance for a multi-centre
thermography study. Thermology International 2018, 18(1): 7–13.
Martin MJ Knazovicka L , McEvoy H , Machin G , Pusnik I , Cardenas D , Sadli M , Chengdu
B , Li W , Saunders P , Girard F Best practice guide: Use of infrared forehead thermometers
to perform traceable non-contact measurements of human body temperature. Accessed on 2
December 2021.
[Link]
NFRARED_FOREHEAD_THERMOMETERS_TO_PERFORM_TRACEABLE_NON-
CONTACT_MEASUREMENTS_OF_HUMAN_BODY_TEMPERATURENg
Ng EYK , Kawb GJL , Chang WM Analysis of IR thermal imager for mass blind fever
screening. Microvascular Research 2004, 68(2): 104–109.
[Link]
Niehues T The febrile child: diagnosis and treatment. Deutsches Arzteblatt International
2013, 110(45): 764–774.
Perpetuini D , Filippini C , Cardone D , Merla A An overview of thermal infrared imaging-
based screenings during pandemic emergencies. International Journal of Environmental
Research and Public Health 2021, 18(6): 1–12. [Link]
Piccinini F , Martinelli G , Carbonaro A Reliability of body temperature measurements
obtained with contactless infrared point thermometers commonly used during the COVID-19
pandemic. Sensors 2021, 21(11): 3794. [Link]
Ring EFJ , Jung A , Kalicki B , Zuber J , Rustecka A , Vardasca R New standards for fever
screening with thermal imaging systems. Journal of Mechanics in Medicine and Biology 2013,
13(2). [Link]
Romanovsky AA , Almeida MC , Aronoff DM , Ivanov AI , Konsman JP , Steiner AA , Turek
VF Fever and hypothermia in systemic inflammation: Recent discoveries and revisions. Front
Bioscience 2005, 10(1–3): 2193–2216.
Roth J , Blatteis CM Mechanisms of fever production and lysis: lessons from experimental
LPS fever. Comprehensive Physiology 2014, 4: 1563–1604.
Shu C , Ding X , Fang C Histogram of the oriented gradient for face recognition. Tsinghua
Science and Technology 2011, 16(2): 216–224. [Link]
0214(11)70032-3
Umapathy S , Krishnan PT Automated detection of orofacial pain from thermograms using
machine learning and deep learning approaches. Expert Systems, 2021, 8(7), e12747.
[Link]
Viola P , Jones, M Rapid Object Detection Using a Boosted Cascade of Simple Features.
Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and
Pattern Recognition. CVPR 2001 n.d., 1:I-511–I-518. IEEE Computer Socilogy.
[Link]
Waalen J , Buxbaum JN Is older colder or colder older? The association of age with body
temperature in 18,630 individuals. Journal of Gerontol A Biology Science Medicne Science
2011, 66A(5): 487–492.
Wald NJ Guidance on terminology. Journal of Medicine Screen 1994, 1: 76.
Wang Y-Q An analysis of the viola-jones face detection algorithm. Image Processing On Line
2014, 4(June): 128–148. [Link]
Wright WF , Auwaerter PG Fever and fever of unknown origin: Review, recent advances, and
lingering dogma. Open Forum Infectious Diseases 2020, 7(5): 132.
Wunderlich CA On the Temperature in Diseases. A Manual of Medical Thermometry.
Translated from the Second German Edition by W Bathurst , MD Woodman , The New
Sydenham Society, London, 1871.
Yang S , Luo P , Loy C-C , Tang X From Facial Parts Responses to Face Detection: A Deep
Learning Approach. 2015 IEEE International Conference on Computer Vision (ICCV), IEEE,
2015, pp. 3676–3684. [Link]
Zhang J , Liu S , Zhu B Fever screening methods in public places during the COVID-19
pandemic. Journal of Hospital Infection 2021, 109(March): 123–124.
[Link]
Zhang K , Zhang Z , Li Z , Qiao Y Joint face detection and alignment using multitask
cascaded convolutional networks” IEEE Signal Processing Letters 2016, 23(10): 1499–1503.
[Link]
Zhou Y , Ghassemi P , Chen M , McBride D , Casamento JP , Joshua Pfefer T , Wang Q
Clinical evaluation of fever-screening thermography: Impact of consensus guidelines and
facial measurement location. Journal of Biomedical Optics 2020, 25(09): 97002–97003.
[Link]

Ethical Aspects in Thermal Imaging Research

Center for Devices, and Radiological Health . 2021. Thermal imaging (infrared thermographic
systems/thermal cameras). U.S. Food and Drug Administration. Retrieved March 5, 2021 ,
from [Link]
imaging-systems-infrared-thermographic-systems-thermal-imaging-cameras
Center for Devices, and Radiological Health . n.d. “Medical Devices.” U.S. Food and Drug
Administration. FDA. Accessed July 6, 2022 . [Link]
Ghassemi P , Pfefer TJ , Casamento JP , Simpson R , Wang Q . Best practices for
standardized performance testing of infrared thermographs intended for fever screening.
PLoS One 2018, 13(9): e0203302. [Link]
Indian Council of Medical Research. 2017. National ethical guidelines for biomedical and
health research involving human participants. Retrieved January 10, 2022 , from
[Link]
International Council for Harmonisation (ICH). 2021. Ich e6 guideline for good clinical practice
(gcp)-update on progress public web conference report. Retrieved January 10, 2022 . from
[Link]
[Link]
Ring EF , Ammer K . Infrared thermal imaging in medicine. Physiological Measurement 2012,
33(3): R33–R46. [Link]
WHO. 2016. Handbook for good clinical research practice (gcp). Retrieved January 10, 2022
, from [Link]