Quantitative T2 mapping detects pathology in normal-appearing brain regions of relapsing-remitting MS patients
Timothy Shepherd1,2, Ivan Kirov1,2, James S Babb1,2, Mary T Bruno2, Robert E Charlson3, Jacqueline Smith2, KAI Tobias Block1,2, Daniel K Sodickson1,2, and Noam Ben-Eliezer1,2

1Center for Advanced Imaging Innovation and Research (CAI2R), New York University School of Medicine, New York, NY, United States, 2The Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York University School of Medicine, New York, NY, United States, 3Department of Radiology, New York University School of Medicine, New York, NY, United States


Accurate quantification of T2 values in vivo is a long-standing challenge hampered by the inherent inaccuracy associated with rapid multi-SE sequences. This inaccuracy is, moreover, not constant and depends on both the pulse sequence scheme and parameter-set employed, resulting in different vendors or scanners yielding different results! We used a recently developed novel T2 mapping technique, the EMC algorithm, to quantify T2 changes in different brain regions of MS patients. Our results demonstrate that the robustness of the EMC approach allows the detection of subtle, but statistically significant T2 differences in normal appearing brain regions for MS patients.


Multiple sclerosis is a chronic disease of inflammation and neurodegeneration. Conventional T2-weighted MRI can detect most areas of active inflammation and white matter (WM) regions with substantial myelin pathology, but visible lesion burden does not always correlate well with patient disability, disease progression, or many symptoms such as cognitive impairment1,2. Gray matter (GM) regions are known to be involved in MS from an early stage, yet also seldom show signal abnormalities on conventional MRI. There remains a strong need for new, more sensitive biomarkers of occult MS brain pathology for different stages or variations of the disease. We recently described a novel, highly accurate method – the echo-modulation curve (EMC) technique – allowing to quantify T2 relaxation values from multiple spin echo (MSE) acquisitions in a manner that is independent of the specific MRI sequence, parameter-set, and scanner being used3,4. In this study, we demonstrate the potential for EMC-based T2 mapping to detect occult pathology in a large cohort of patients with relapsing-remitting multiple sclerosis (RR-MS).


Data acquisition: 39 healthy volunteers (23 males) and 27 patients (9 males) with clinically diagnosed RR-MS were imaged on a whole-body 3T scanner (Siemens Skyra) using a T2 mapping MSE protocol and a T1 weighted reference. Scan parameters were {TR=2500 ms, Echo-spacing=12 ms, Nechoes=10, res=1.7x1.7mm2, slice=3 mm, bandwidth=200 [Hz/Px], Tacq=2:44min using 2x GRAPPA acceleration}.

EMC algorithm: Bloch simulations of the prospective MSE protocol were performed using the exact RF pulse shapes and other experimental parameters. Simulations were repeated for a range of T2 and B1+ values (T2=1…1000ms, B1+ = 50…130 % deviation from nominal value), producing a database of EMCs, each associated with a unique [B1+,T2] value pair. T2 maps were then generated by pixel-by-pixel matching of the MSE DICOMs time-series, to the database of simulated EMC via l2-norm minimization of the difference between experimental and simulated EMCs3.

Segmentation: Freesurfer5 was used to delineate 7 regions-of-interest (ROIs) in the T2 maps based on a co-registered volumetric T1 dataset. Separately, manual ROIs were drawn for 14 GM and WM structures by a board certified neuroradiologist on T2-weighted images, synthetically generated based on the EMC fitted T2 map (Figure 1).

Statistical analysis: Mean and standard deviation (SD) were calculated for each ROI. Analysis of covariance (ANCOVA) was used to test the statistical difference between MS and control groups in terms of T2 mean and SD for each ROI, adjusted for age and gender. Logistic regression and area under the ROC curve (AUC) was used to assess the utility of mean T2 for discriminating MS patients from controls. Region-specific T2 values were also correlated with disease duration and patient-reported disability6 (ordinal scale of 0-8 with 8 being bedridden).


Tables 1 & 2 summarize the average T2 values for the Freesurfer and manually drawn brain ROIs respectively. Statistically significant T2 differences were detected in the MS versus healthy controls groups in 4 out of 7 Freesurfer ROIs, and 9 out of 14 for manually drawn ROIs. AUC analysis showed that thalamic T2 changes could discriminate RR-MS from control subjects (AUC = 0.913). Last, whole-brain WM T2 positively correlated with the patient reported disability scale (p-value 0.038). Full benchmark tests of the EMC technique were also run across different scanners and parameter sets (results not shown) yielding high reproducibility (mean SD=1.8% across 24 scans) and low inter-subject and inter-scanner variability (mean SD=2.4 % across 30 volunteers).


A novel, highly accurate EMC technique can detect subtle T2 differences for a variety of normal-appearing brain structures in a large cohort of RR-MS patients compared to healthy controls. Significant T2 changes were observed for the thalamus and other GM regions (see Table 2). GM injury can occur in earliest stages of MS and is associated with a wide range of clinical symptoms including cognitive decline, motor deficits, fatigue, and pain syndromes7. As expected, WM lesions visible on conventional MRI had T2 values 37-45% higher than homologous WM regions in healthy controls (all comparisons, p-value < 0.001). Also noted, were significant T2 increases for the manually drawn ROIs in the normal-appearing WM regions such as the periventricular WM, genu, and body of the corpus callosum. The biophysical basis for these underlying T2 changes will be complex and may reflect different pathological aspects of MS including GM demyelination, microglial activation, iron deposition and/or decreased neuronal density. Overall the current results suggest EMC-derived T2 mapping could become a useful imaging biomarker of occult MS pathology in normal appearing brain structures. Future efforts will compare regional T2 changes across time and between MS disease classifications.


Financial support: Helen and Martin Kimmel Award for Innovative Investigation. NIH Grants: P41 EB017183; RO1 EB000447.


[1] Hauser SL, Oksenberg JR. The neurobiology of multiple sclerosis: genes, inflammation, and neurodegeneration. Neuron. 2006; 52(1): 61-76.

[2] Filippi M et al. Ultra-high-field MR imaging in multiple sclerosis. J Neurol Neurosurg Psychiatry. 2014; 85(1):60-6.

[3] Ben-Eliezer N, Sodickon, DK, and Block, KT. Rapid and accurate T2 mapping from multi-spin-echo data using Bloch-simulation-based reconstruction. Magn Reson Med 2015; 73(2): 809-17.

[4] Ben-Eliezer N1, Sodickson DK, Shepherd T, Wiggins GC, Block KT. Accelerated and motion-robust in vivo T2 mapping from radially undersampled data using Bloch-simulation-based iterative reconstruction. Magn Reson Med 2015; doi: 10.1002/mrm.25558. [Epub ahead of print].

[5] Fischl, B. et al, 2002. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron 33:341-355.

[6] Vollmer TL, Ni W, Stanton S, Hadjimichael O. The NARCOMS patient registry: a resource for investigators. Int J MS Care 1999;1:12-15.

[7] A Minagar, MH Barnett, RHB Benedict, D Pelletier, I Priko, MA Sahraian, E Frohman, R Zivadinov. The thalamus and multiple sclerosis – Modern views on pathologic, imaging and clinical aspects. Neurology 2013; 80:210-219.


Figure 1: Two representative slices showing examples of manual brain segmentation drawn on T2 weighted images that were synthetically produced using the EMC technique. (a) Optic radiations [1,2], temporal stem [3,4], globus pallidus [5,6]. (b) Genu [1] and splenium [2] of corpus callosum.

Table 1: EMC based T2 values (mean ± SD) for 39 healthy controls vs. normal appearing tissue in 27 RR-MS patients, over 7 brain regions segmented using Freesurfer. Statistically significant separation (p-value < 0.05) is seen between healthy and normal-appearing tissues in 4 out of 7 ROIs.

Table 2: EMC based T2 values (mean ± SD) for 39 healthy controls vs. normal appearing tissue in 27 RR-MS patients, over 14 brain regions manually segmented by a board-certified radiologist. Statistically significant separation (p-value < 0.05) is seen between healthy and normal-appearing tissues in 9 out of 14 ROIs.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)