MAGNETOM Flash (71) 2/2018 How-I-do-it

MR Spectroscopy in Neuroimaging –
A Practical Guide to Integrate a Complex
Technology into a Clinical Workflow
Marco Essig, M.D., Ph.D., FRCPC1; Craig Snell, RTMR, RTR1; Lawrence Ryner, Ph.D.1,2

1
Department of Radiology, Health Sciences Center Winnipeg, Canada
2
Department of Physics and Astronomy, University of Manitoba, Winnipeg, Canada

Introduction benign from malignant features. Specific character-

ization facilitates planning of the most appropriate
After the introduction of NMR spectroscopy in chemistry treatment. Furthermore, functional neuroimaging
laboratories in the 1950s, it took decades for MR spec- of central nervous system (CNS) neoplasms can be
troscopy to eventually be used in a clinical neuroimaging expanded to the monitoring of ongoing therapy and for
environment. Even as late as 1996, Mauricio Castillo early detection of therapeutic side effects. The predictive
noted that magnetic resonance spectroscopy (MRS) assessment of therapy response and the monitoring of
had received little attention from the clinical radiology ongoing therapy to guide therapeutic intervention are
community. Indeed, most MR spectroscopic studies major challenges in the current treatment of CNS
were initially performed by a small and dedicated group neoplasms.
of individuals, mostly basic scientists, partly because Proton magnetic resonance spectroscopy is one
MR spectroscopy did not produce “pictures” but resulted of the most important functional imaging techniques
in “graphs” called “spectra”, and also because of long and provides detailed tumor information beyond the
acquisition times. Much has been done over the years morphologic characteristics observed in standard images.
to make the acquisition and processing of spectroscopy Single Voxel Spectroscopy (SVS) is used to assess a single
data easy, faster, and user friendly, but, still, the full voxel covering the brain lesion or for larger lesions just
integration of MR spectroscopy seems a challenge for a selection of the lesion. Chemical Shift Imaging (CSI),
many clinical Radiology departments. also known as MR spectroscopic imaging (MRSI), is used
In this article, we describe how MR spectroscopy to assess multiple voxels within and surrounding the
and MR spectroscopic imaging can be integrated into the tumor. Both methods have become valuable clinical
clinical workflow of a busy, high-volume MRI department tools, especially in the diagnostic work-up of tumors
using the example of brain tumor imaging. and the differentiation of normal from pathological
conditions.
MRS for brain tumor imaging 1
H protons are used most commonly in clinical
The goals and requirements for brain tumor imaging practice because of the abundance of 1H protons in
are making a diagnosis and/or a differential diagnosis, body tissue and the consequently high signal-to-noise
while accurately grading and delineating lesions for ratio, although it is possible to acquire signal from
tumor description and risk assessment. Imaging is also phosphorous-31, carbon-13 and other nuclei. The
involved in the decision-making process for treatment spectroscopic characterization of brain abnormalities
such as precise planning of surgical or radiotherapeutic relies mostly on the calculations of ratios between the
interventions both of which require an optimal lesion major proton metabolites including N-acetylaspartate
detection and delineation. Following therapy, neuroim- (NAA – a neuronal marker), choline-containing com-
aging techniques have been shown to be very important pounds (Cho – a marker of membrane turnover), and
for monitoring of disease and to identify and monitor creatine-phosphocreatine (Cr – important in the tissue
possible therapy related side effects. cell energy cycle), and on the presence of lactate
In recent years, functional neuroimaging techniques (Lac – a marker of anaerobic glycolysis), and lipids [1-6].
were implemented to optimize tumor characterization, Brain tumors typically have a loss of NAA and an increase
with an emphasis on improved specificity to separate of the Cho content (Fig. 1).

siemens.com/magnetom-world 73
How-I-do-it MAGNETOM Flash (71) 2/2018

1 2

Figure 1: Brain tumor spectrum showing increased Cho and Figure 2: CSI example showing the spatial distribution of
reduced NAA. the Cho/NAA ratio.

MR spectroscopy has been used to differentiate non- planning of biopsy and for the treatment decision and
tumor lesions, like harmatomas from gliomas [7-8]. In management, it is essential to identify the highest tumor
several studies, it has been reported that harmatomas grade in those often very morphologically homogeneous
did not differ significantly from the normal brain or appearing tumors. The authors were also able to provide
physiological changes, while gliomas had lower NAA/Cr, thresholds for the metabolite ratios for the diagnosis of a
Cr/Cho, and NAA/Cho ratios. In patients with seizures it high-grade tumor. A review of the literature, taking into
is important to identify tumor changes from scar as this account differences in MR spectroscopic technique such
has a major impact on the further management of as the choice of echo time (TE) and method for determi-
patients. In a study by Vuori et al. [6], patients experi- nation of metabolite ratios, demonstrated that the mean
encing seizures who also had a cortical brain lesion on maximal values obtained for Cho/Cr and Cho/NAA and
MR images were studied with proton MR spectroscopy. mean minimum values for NAA/Cr ratios in their study
A metabolite ratio analysis was performed, and the (1,7) were comparable to previously published data in
metabolite signals in the lesion core were compared with differentiating between low- and high-grade gliomas.
those in the contralateral centrum semiovale and in the As in many functional techniques, one of the
corresponding brain sites of control subjects to separate- challenges is the use of standardized data acquisition
ly obtain the changes in NAA, Cho, and Cr. In their study, and post-processing techniques. For data acquisition,
ten patients had a low-grade glioma (three, oligodendro- the same echo time (TE) should be used, and data from
gliomas; three, oligoastrocytomas; three, astrocytomas; the contralateral unaffected brain tissue should be
and one, pilocytic astrocytoma), and eight had FCDM acquired. Studies comparing long and short TE acquisi-
(five, focal cortical dysplasias and three, dysembryoplas- tion found that a short TE provided a slightly better
tic neuroepithelial tumors). They found that loss of NAA tumor classification [10-11]. In a study by Majos et al.,
and increase of Cho were more pronounced in low-grade long TE acquisitions were only beneficial in studies of
gliomas than in subjects with cortical developmental meningiomas [12].
malformations. MR spectroscopy was also able to The development in modern scanner technology
differentiate between subtypes of gliomas. further allows for the simultaneous measurement of
In a study by Law et al. [9], MR spectroscopy and MR spectroscopic data from more than just a single voxel.
perfusion enabled identification at a high sensitivity and This technique is called chemical shift imaging (CSI) or
a high positive predictive value for tumor grading when magnetic resonance spectroscopic imaging (MRSI) and
compared with conventional contrast enhanced MR enables the acquisition of multiple small voxels in two or
imaging. It is well known that there are a certain percent- even three dimensions, providing better information
age of high-grade tumors that do not present with a about the spatial heterogeneity of a lesion (Fig. 2). The
blood-brain-barrier disruption or vice versa. For the voxel information can be used to calculate metabolite

74 siemens.com/magnetom-world
MAGNETOM Flash (71) 2/2018 How-I-do-it

ratios, which can then be color-coded and superimposed How to – MR Spectroscopy

on the anatomic images to better visualize hot spots
within a tumor. Getting started
Follow-up assessment of cerebral tumors is a Before discussing the details of processing MR spectros-
promising field for MR spectroscopy. Increases in size copy data using syngo.via, it is important to ensure that
and contrast enhancement are typical findings in tumor the acquisition of the spectroscopy data is optimized.
progression but also reflect therapeutic-induced changes This section provides some tips to assist in acquiring the
and/or postoperative changes. The ratio of choline to single voxel and chemical shift imaging data.
normal creatine level is usually significantly elevated in As mentioned in the previous section, when setting
those areas consistent with tumor compared to those up the CSI sequence selecting a lower TE (i.e., 30 ms)
containing predominantly treatment effect. In fact, will yield better spectral quality in terms of tumor
treatment effect is generally indicated by a marked classification. Typically, the CSI sequence should be
depression of all the intracellular metabolite peaks placed after the initial imaging sequences. This allows
from choline, creatine, and N-acetyl compounds. for the acquisition of imaging data in each plane so
MR spectroscopy alone may not be helpful in in- that the spectroscopy excitation voxel can be accurately
stances where patients have mixed histologic findings placed on the contrast-enhancing or most suspicious
comprised of necrosis and tumor. Because of this tumor components. There are several important scan-
heterogeneity and as a result of low spatial resolution, ning parameters to keep in mind (Fig. 3):
MRS findings of choline and NAA resonances below the 1. Table position, found in the menu under System/
normal range may indicate variable histologic findings Miscellaneous/Position Mode, should be “REF mode”.
ranging from radiation necrosis, gliosis, and macrophage If ISO or FIXED mode is used, the isocenter will move
infiltration to mixed tissues that contain some regions of and syngo.via will not be able to triangulate where
tumor. The careful choice of voxel placement and inter- the voxel was placed on the images with different
pretation of results in concordance with other imaging ISO centers.
and clinical findings are critical in distinguishing between 2. When the voxel is placed, click on the scroll header
tumor and treatment-related changes. Furthermore, and select nearest, confirm placement on all three
validation studies using image-guided tissue biopsy planes, and confirm that the voxel is not encroaching
need to be performed to correlate imaging with histo- on any extraneous tissues.
logic findings.

3
Figure 3: Setting up the CSI
acquisition.

siemens.com/magnetom-world 75
How-I-do-it MAGNETOM Flash (71) 2/2018

3. 2D Distortion Correction must be turned off when Locating data on syngo.via

acquiring an imaging localizer for spectroscopy. Once logged into syngo.via, the data should be located.
If this is left on, the voxel location may be affected There are two ways to do this; send the dataset directly
by the amount of distortion correction applied to from the modality to the syngo.via server or query/
the images. Note that distortion correction can be retrieve the data from PACS on syngo.via. Configuration
retroactively reversed so that images can still be used of a PACS node by the local service engineer will be
for voxel positioning. required when sending from the modality (AE title,
Placing a voxel, whether for single voxel or CSI, will IP address, and port number will need to be provided).
follow similar principles: Avoid including any lipid signal
within the excitation voxel as this will cause a large lipid Query/Retrieve PACS
peak to dominate the spectrum; stay away from the skull Searching for cases on PACS is done in the DICOM
as the bone marrow will also result in an overwhelming retrieve header in the bottom third of the home screen.
lipid peak; and try to stay from the sinuses where it Identify the Source selection which is the local PACS
might be difficult to optimize the magnetic field homo- server the data is stored on. The search filters header is
geneity, i.e., the shim. The voxel in Single Voxel Spectros- where the patient demographics will be put in. Click the
copy (SVS) is usually small enough (20 x 20 x 20 mm3) search button and observe all cases on PACS meeting the
that it can be accurately placed to avoid such extraneous search criteria. Find the desired patient exam and select
tissues. Take care that the SVS voxel size is not set too the retrieve icon (Fig. 4).
small as this will result in very noisy spectra. CSI will
require placement of several saturation bands to prevent Browser search
contamination from tissues outside of the volume-of- Now that the data has been downloaded on to the
interest (VOI). Saturation bands in all three planes are server, search for it in the patient browser. In the top
extremely helpful in eliminating these signals. Take care half of the patient browser, type the name in and search
that the CSI voxel size does not get too small, as this will for the data (Fig. 5). Search the database manually if the
result in noisy metabolite images. search function is not working.
MR spectroscopy is extremely sensitive to any Now that the dataset has been located, open it
metal that might cause magnetic field inhomogeneities. with the MR Spectro Analysis program by right-clicking
Items such as metal dentures, belt buckles, shoes, metal on the patient in the browser, select OPEN WITH, and
clothing snaps, removable body piercings, etc., should be scroll down to find “MR SPECTRO”. Selecting the star
removed from the patient prior to the scan (standard MRI icon beside “MR Spectro” will identify this workflow as
practice). In addition, the closer the voxel is to isocenter, a favorite and make it easily accessible for subsequent
the better the shim will be, as measured with the Full analyses.
Width Half Maximum (FWHM). The FWHM indicates
how broad or narrow the peaks in the spectrum will be. Selecting the protocol
Typically for SVS the FWHM should be under 12 Hz and After selecting the MR Spectro workflow, wait for the
for 2D multi-voxel under 15 Hz, although these values data to load, process, and display on the screen. This
are for 1.5T and can vary significantly depending on VOI might take ten seconds or so. When MR SPECTRO opens,
location and size.

4
Figure 4: Query/Retrieve
window

5
Figure 5: Browser search

76 siemens.com/magnetom-world
MAGNETOM Flash (71) 2/2018 How-I-do-it

multiple windows appear showing the spectral grid Furthermore, there are many different configurations
overlay on the anatomical image, the spectrum for the that can be used for viewing. Simply click the “4 win-
selected voxel, the orthogonal image planes with the dows” icon located just to the right of MR Spectro
spectral grid overlay, and the color metabolite map. Analysis label (Fig. 7) and select the desired template.
The complete list of data series can be viewed by clicking The next step is only required if you wish to add
the small arrow located to the side. The series that are metabolites to the analysis. You should only have to
highlighted in blue in the Data Groups window are the do this once – syngo.via will remember the workflow
ones that have been automatically selected by the protocol. To select the metabolites that are to be
program (Fig. 6). Different series can be dragged and analyzed, click the Spectro browser icon (Fig. 8) found
dropped into any window as long as the requirements at the bottom of the MR Spectro Analysis window.
mentioned in “Getting started” were followed. Every The following example shows how to add the lactate
window will have tools in each corner to alter that metabolite to the analysis. Highlight csi_se_30 (Fig. 9)
specific window if needed. Note that the default screen and select edit. A new window will appear. Select
window configuration will be different depending on the “Prior Knowledge” and click the + icon (Fig. 10). Another
monitor resolution and size. window will appear – this is where all the metabolites
are listed. Scroll down, highlight lac_se_30, and click the
select icon (Fig. 11). This loads the lactate metabolite
acquired with a spin echo sequence at an echo time of
6 30 ms into the analysis. Save the protocol with a new
name. Select “Save as” and save the new protocol as
“csi_se_30_lac”.
The last step is to select the new protocol. Right
click on the series in the MR Spectro window and click
on “Select Protocol”. Choose the new protocol that was
just created (Fig. 12). The program will automatically
reprocess the data and queue it at the bottom of the
Spectro workflow.

Defining the result table

The next step is to modify the Result Table to select the
desired metabolites and metabolite. Select myo-inositol
(mIns), NAA, Cho, Cr1, Cr2, and Lac. Creatine is often
used as a baseline in spectral analysis since its concentra-
tion stays fairly constant. The choline to NAA ratio (Cho/
NAA) is another common ratio used in assessing brain
tumor spectra, since the choline peak is often elevated
and the NAA peak is often reduced. In the spectrum
window, click on the ruler in the top right corner. This
will be where the result table is defined (see Fig. 13). In
the pop-up window, there are useful examples of how to
Figure 6: Data Groups enter the desired equations. Using creatine as the base,
defined values will display when adding the Integrals.

7 8

Figure 7: Configuration icon Figure 8: Spectro browser icon

siemens.com/magnetom-world 77
How-I-do-it MAGNETOM Flash (71) 2/2018

9 10

Figure 9: Editing the csi_se_30 protocol Figure 10: Entering prior knowledge

11 12
Figure 11:
Selecting the
lac_se_30 protocol

Figure 12: Selecting the

analysis protocol

13

Figure 13: Defining the result table

78 siemens.com/magnetom-world
MAGNETOM Flash (71) 2/2018 How-I-do-it

The first column in the table is the label of the metabo- 14

lite ratio. The second column is the actual value of that
ratio. The third column is the percent variation which is
a measure of how good the spectral fit was. A rule-of-
thumb is that metabolite ratios with percent variations
greater than 20% should not be considered accurate.
The lower this number is, the better the fit.
The result table can be manipulated in many ways;
adding and subtracting different combinations of ratios
will yield results tailored to specific differential diagno-
ses. After the result table is properly defined, click “OK” Figure 14: Defining the metabolite image
and syngo.via will remember these values for all subse-
quent analyses when the protocol is saved.
will take a screen shot of every window that is highlight-
Defining metabolite maps ed blue. Layout Snapshot will take a screen shot of all 4,
Metabolite images or maps are another way of visualiz- or 6 windows, depending on the window configuration.
ing the spatial variation of metabolites across the brain, “Snapshot to finding” will add the selected windows into
and for seeing “hot spots” for certain metabolites and the chosen findings folder that is identified by a check-
ratios. The “Define Metabolite Image” window (Fig. 14) mark. A new findings folder is created for each new snap-
is used to define which maps to display, using equations shot; therefore once the initial snapshot is done, use the
similar to the way the result table was defined. Metabo- “Snapshot to finding” icon. These preceding snapshots
lite map windows are typically on the right-hand side of to findings will be archived into the findings folder that
the monitor. First, drag and drop the image-underlay were selected. Each findings folder should be exported to
series into the window from the Data Groups (Fig. 6). PACS in a different series. Review all snapshots made in
Note that the available series are displayed as “Data the findings navigator by double clicking on the small
groups” by default, but can be viewed as “Thumbnails”, picture icon. When all the snapshots are done, right click
or “Study/Series” by selecting the dropdown. Any image on the findings and select export to PACS (Fig. 16). For
series can be chosen (usually axial T1 post gad) as long CSI, ensure that the full voxel position is included in the
as the series was scanned under the specifications snapshot, as in Fig. 17 and 18.
mentioned in getting started, i.e., same isocenter, no
2D distortion correction, etc. Once the series is loaded Healthy vs. unhealthy analysis
in the window, select the ruler (define metabolite image) It can be useful to select two spectra from the full CSI
in the upper right-hand corner and define any type of dataset, one corresponding to the spectrum with the
color map that is required (Fig. 15) Whether they are most abnormal spectral ratios (labeled “Unhealthy”)
single metabolite maps, or ratios, use the same method and another from contralateral normal appearing brain
as was used for defining the result table. tissue (labeled “Healthy”). Once the syngo.via spectros-
If gaps are seen in the metabolite image, it might copy analysis is configured with all the tables and all the
be because a spectrum has been automatically identified maps defined, select the two representative “Healthy”
as a “Low quality voxel”. This means that the spectral and “Unhealthy” voxels (Figs. 17, 18) and collect a
fitting was unreliable in this voxel and it shouldn’t be snapshot of each for export to PACS for interpretation
used in the diagnosis. It is also possible to manually by the radiologist.
identify voxels as “Low quality”, should this be necessary.
For example, if the metabolite map for lactate shows a
very bright spot, click on that voxel and examine the
Strengths/limitations
spectrum. If a very high lipid signal is observed, then
Overall, the syngo.via MR spectroscopy analysis software
mark that voxel as “Low quality” so that it doesn’t mislead
is extremely user-friendly and easy for technologists,
the radiologist.
radiologists, and other clinical staff to use.
• The modern graphic user interface makes full use
Exporting results to PACS
of multiple windows, context-sensitive mouse-clicks,
Now that the Result Table has been generated and the
and a well thought out workflow.
metabolite images have been defined, the results can be
• A key benefit is that the software package is ready-
sent to PACS for reporting and archiving purposes. Under
to-use, right from the start, and does not require
Tools (bottom third of screen), click the arrow below the
the acquisition of any additional scans on various
“layout snapshot” icon. There are three options. Snapshot
spectroscopy phantoms.

siemens.com/magnetom-world 79
How-I-do-it MAGNETOM Flash (71) 2/2018

15

Figure 15: Examples of metabolite maps

16

Figure 16: Exporting to PACS

17 18

Figure 17: Selecting the voxel for the “Healthy spectrum” Figure 18: Selecting the voxel for the “Unhealthy spectrum”
corresponding to the brain tumor spectrum with the “worst”
spectral ratios

80 siemens.com/magnetom-world
MAGNETOM Flash (71) 2/2018 How-I-do-it

19
Figure 19:
T1, T2, and FLAIR images of a patient
following radiation therapy of a high-grade oligo-
dendroglioma

20 21

Figure 21: Second patient with a glioblastoma multiforme

showing FLAIR and T1-weighted image

• The spectral fitting routine has been fully updated

from previous versions and now utilizes robust
Figure 20: Single voxel spectrum from the suspicious area time-domain fitting which is less dependent on
showing relatively normal Cho to NAA ratio
perfect phasing as is required for spectral domain fits.
• Inclusion of a goodness of fit measure in the
22 Results Table helps in determining which spectral
fits are reliable (e.g., rule of thumb is a variation of
less than 20%).
• Automatic marking of spectra with low quality fits
helps the user determine which fits are reliable and
which aren’t.
Any software package comes with limitations that are
often fixed in subsequent releases, based on customer
feedback.
• The lack of an ability to sum multiple spectra from a
user-selected region of interest in the CSI grid makes
it hard to fully assess larger tumors extending over
multiple voxels.
• Other advanced CSI software packages come with the
ability to shift the CSI grid in order to position voxels
more precisely over suspicious regions in the image;
this ability is lacking in this software package.
It should also be noted that there is an Advanced Mode
Figure 22: Single voxel spectrum from the suspicious area that can be configured in the Workflow Step Configura-
showing relatively normal Cho to NAA ratio, with some tion which provides some additional functionality to the
changes in the lactate/lipid region software. Users wishing to have this enabled should
consult Siemens service.

siemens.com/magnetom-world 81
How-I-do-it MAGNETOM Flash (71) 2/2018

Patient cases Acknowledgements

Two patient cases are presented to demonstrate the The authors thank Jerry Moran, Siemens Canada
utility of MR spectroscopy in helping to differentiate Research Collaborations Manager, and Uwe Boettcher,
treatment-related changes from progressing tumor as Siemens Healthineers Germany, for their assistance in
observed in standard MR imaging. setting up and providing instructions on the use of the
In this first patient case, the standard T1- and syngo.via spectroscopy package.
T2-weighted imaging (Fig. 19) appeared to demonstrate
an increase in tumor size after radiation therapy of the References
high-grade oligodendroglioma.
1 Alger JR, Frank JA, Bizzi A, et al. Metabolism of human gliomas:
Single voxel spectroscopy from the suspicious region
assessment with H-1 MR spectroscopy and F-18 fluorodeoxyglu-
indicated (Fig. 20) showed a relatively normal ratio of cose PET. Radiology 1990; 177:633-641.
Cho to NAA signal without the signature elevation of 2 Negendank WG, Sauter R, Brown TR, et al. Proton magnetic
the Cho to NAA ratio found in brain tumor tissue, thus resonance spectroscopy in patients with glial tumors: a multi-
indicating that the changes are not likely related to center study. J Neurosurg 1996; 84:449-458.
3 Meyerand ME, Pipes JM, Mamourian A, Tosteson TD, Dunn JF.
tumor progression but rather related to treatment
Classification of biopsy-confirmed brain tumors using single-vox-
effects.. el MR spectroscopy. AJNR Am J Neuroradiol 1999; 20:117-123.
A second patient case with glioblastoma multiforme 4 Dowling C, Bollen AW, Noworolski SM, et al. Preoperative proton
shows a suspicious area posterior to the resected region MR spectroscopic imaging of brain tumors: correlation with
following chemotherapy and radiation therapy (Fig. 21). histopathologic analysis of resection specimens.
AJNR Am J Neuroradiol 2001; 22:604-612.
Single voxel spectroscopy from a voxel located in the
5 Demaerel P, Johannik K, van Hecke P, et al. Localized 1H NMR
middle of the suspicious region is shown in Figure 22. spectroscopy in fifty cases of newly diagnosed intracranial
With no increase in the Cho to NAA ratio and some tumors. J Comput Assist Tomogr 1991; 15:67-76.
changes in the lactate/lipid region at 1.3 ppm, these 6 Vuori K, Kankaanranta L, Hakkinen AM et al. Low-grade gliomas
changes in the MR imaging are most likely treatment- and focal cortical developmental malformations: differentiation
with proton MR spectroscopy. Radiology 2004 Mar; 230(3):703-8.
related changes and not a progressing tumor.
7 Wilson M, Cummins CL, Macpherson L, et al. Magnetic resonance
spectroscopy metabolite profiles predict survival in paediatric
Conclusion brain tumours. Eur J Cancer 2013; 13:457-464
8 Norfray JF, Darling C, Byrd S, et al. Short TE proton MRS and
neurofibromatosis type 1 intracranial lesions. J Comput Assist
The syngo.via MR spectroscopy analysis software is a Tomogr 1999; 23:994-1003
powerful, easy-to-use tool that streamlines and simplifies 9 Law M. MR spectroscopy of brain tumors.
the use of MRS and MRSI as a routine functional imaging Top Magn Reson Imaging 2004 Oct; 15(5):291-313.
10 Delorme S, Weber MA. Applications of MRS in the evaluation of
method in a high-volume MRI department. Using a stan-
focal malignant brain lesions. Cancer Imaging 2006 Jun 22; 6:95-9.
dardized acquisition and analysis workflow, as outlined 11 Galanaud D, Nicoli F, Chinot O, Confort-Gouny S, Figarella-Brang-
in this article, enables the use of this complex technology er D, Roche P, Fuentes S, Le Fur Y, Ranjeva JP, Cozzone PJ. Nonin-
in a quick, easy-to-implement process on a day-to-day vasive diagnostic assessment of brain tumors using combined in
routine basis that provides key information to the radiol- vivo MR imaging and spectroscopy. Magn Reson Med 2006 Jun;
55(6):1236-45.
ogist. This functional information can be used for better
12 Majos C, Julia-Sape M, Alonso J, Serrallonga M, Aguilera C,
workup and differential diagnosis of brain tumors, for Acebes JJ, Arus C, Gili J. Brain tumor classification by proton MR
treatment planning, and for differentiation between spectroscopy: comparison of diagnostic accuracy at short and
tumor progression and pseudo effects following treatment. long TE. AJNR Am J Neuroradiol 2004 Nov-Dec; 25(10):1696-704.

Contact
Lawrence Ryner, Ph.D. Marco Essig, M.D., Ph.D., FRCPC
University of Manitoba Professor and Chair,
Dept. of Radiology, Department of Radiology
Medical Physics Medical Director of Diagnostic
ON-3 Cancer Care Manitoba Imaging, WRHA
820 Sherbrook Street GA-216, General Centre – HSC
Winnipeg, Manitoba, MB R3T 820 Sherbrook Street
2N2 Winnipeg, Manitoba, MB R3T 2N2
Canada Canada
Tel.: +1 204-787-1400 Tel.: +1 204-787-1335
[email protected] [email protected] Lawrence Ryner Marco Essig Craig Snell

82 siemens.com/magnetom-world