Click here to


Are you sure ?

Yes, do it No, cancel

Comparison of Vendor Provided Analysis Vs Computational Sequence Modeling for T2 Mapping

K McPhee*, A Wilman, Department of Biomedical Engineering, University of Alberta, Edmonton, Alberta


(Sunday, 7/14/2019) 1:00 PM - 2:00 PM

Room: 221CD

Purpose: To examine differences between T2 maps computed using a vendor provided package, and a fit employing a computational model which fully simulates the pulse sequence. Full modeling approaches are commonly employed in research studies to account for inherent imperfections in multi-echo spin echo (MESE) sequences, however, single exponential fitting and vendor provided analysis remain persistent in clinical research. These methods have not previously been directly compared.

Methods: Human brain images were acquired on eight volunteers (mean age 29, range 23-34, 4 male) on a 3T Siemens Prisma MRI system. MESE (TE=10-200ms with 10ms spacing; TR=5s; refocussing angle=180°; voxel size=1.14x1.14x3mm3; matrix=192x156x1; TA 5:32) data was acquired, and T2 maps were computed by Siemens MapIt™. Flip angle maps (B1) were collected on all subjects using a Bloch-Siegert approach (TE=2.24ms; TR=16s; voxel size=1.1x1.1x3.0mm; TA=33s). Additional T2 maps were computed offline using a single exponential fit and using a fully simulated MESE sequence. Simulations of slice-selective RF pulses were performed using the Shinnar-Le Roux algorithm, and relaxation between pulses was calculated according to Bloch equations. Echo time, RF pulse shapes, gradients and spoiling were matched to experimental imaging parameters. Fitting was performed using custom in-house MATLAB code. T2 maps and ROI averages were compared.

Results: Stimulated echoes resulting from B1 inhomogeneity and slice profile contaminate the acquired decay curves. MapIt™ provided similar results to exponential fitting. Both overestimated by up to 52% in examined regions T2 relative to full sequence modelling. B1 varied up to approximately ±20% across the brain. Areas of high and low B1 correspond visually to hot spots in the MapIt™ results.

Conclusion: MapItâ„¢ provides similar results to a single exponential fit, and does not account for indirect and stimulated echoes, which contaminate the acquired decay curves. As a result, T2 values are overestimated.

Funding Support, Disclosures, and Conflict of Interest: Grant support was provided by Canadian Institutes of Health Research, and Natural Sciences and Engineering Council of Canada (NSERC). Salary support for KCM was provided by scholarships from NSERC, and Alberta Innovates Health Solutions.


Relaxation Times, Quantitative Imaging, MRI


IM- MRI : Quantitative imaging/analysis

Contact Email