Improved Myelin Water Imaging Using 3D GRASE and Orthogonal Matching Pursuit Reconstruction
R Gabr*, J Lincoln, K Hasan, J Wolinsky, P Narayana, University of Texas Health Sci. Center at Houston, Houston, TX
Presentations
MO-IePD-TRACK 2-5 (Monday, 7/26/2021) 3:00 PM - 3:30 PM [Eastern Time (GMT-4)]
Purpose: Noninvasive MRI-based quantification of myelin provides valuable information for the study of neurological and neuropsychiatric disorders. Multi-echo spin echo MRI data are analyzed using multi-exponential T2 relaxation to assess myelin by computing the fraction of myelin water (usually assumed as 10
Methods: 3D GRASE data were acquired from one patient with multiple sclerosis in four different session in a study of oral acetazolamide over a period of 5.5 month using a 3.0 T Philips MRI system with a multi-channel head coil. Acquisition parameters were TR/TE/ΔTE = 747/7/7 ms; number of echoes, 32. Non-negative least squares reconstruction was used as reference. Orthogonal matching pursuit was implemented with nonnegativity constraint (NNOMP). The number of components (sparsity) was empirically optimized and four components was found to yield the best results.
Results: The NNOMP-reconstructed MWF maps consistently showed improved visual quality in all scans. Figure 1 shows a comparison between NNLS and NNOMP in one scan. White matter MWF values calculated from NNOMP were approximately 10-15% higher than those obtained from NNLS.
Conclusion: NNOMP produces MWF maps with improved quality. The improved performance is attributed to the regularization imposed by the sparsity constraints in NNOMP. No spatial regularization was imposed on either of the techniques. Analysis on a larger sample is underway to assess NNOMP both qualitatively and quantitatively.
ePosters
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email