Lohith Kini1, Matthias Gebhardt2,
Khaldoun Makhoul3,
1EECS, Massachusetts Institute of
Technology, Cambridge, MA, United States; 2Siemens Medical
Solutions, Erlangen, Germany; 3MGH, Harvard Medical School, A. A.
Martinos Center for Biomedical Imaging, Charlestown, MA, United States; 4Harvard-MIT
Division of Health Sciences and Technology, MIT, Cambridge, MA, United States
Specific
Absorption Rate (SAR) is a major concern in parallel transmission.
Calculating the average N-gram SAR arising at each spatial location is a
computationally intractable problem. We propose and demonstrate improvements
to a recent algorithm, the fast region-growth algorithm, and exploit its
parallelizable computation structure by implementing it on a commercial
graphics card. This algorithm is up to 4-7 times faster in runtime regardless
of model resolution and number of RF pulse time samples while maintaining
accuracy with the fast region-growth algorithm.