Meeting Banner
Abstract #0105

Clinical Performance & Validation of a Compressed Sensing Contrast Enhanced MRI with Fast Image Reconstruction

Shilpy Chowdhury1, Tao Zhang2, Richard A. Barth1, Michael Lustig3, Mark Murphy3, Marcus T. Alley1, Thomas Grafendorfer4, Paul Calderon5, John M. Pauly2, Brian A. Hargreaves1, Shreyas S. Vasanawala1

1Radiology, Stanford University, Stanford, CA, United States; 2Electrical Engineering, Stanford University, Stanford, CA, United States; 3Electrical Engineering and Computer Sciences, University of California Berkeley, Berkeley, CA, United States; 4ATD Coils, GE Healthcare, Stanford, CA, United States; 5MR Hardware Engineering,GE Healthcare, Fremont, CA, United States

Pediatric abdominal MRI is challenged by small anatomical structures and physiologic motion. We assessed performance and clinical validation of a new compressed sensing algorithm in 29 consecutive patients, that permits rapid reconstruction even with high-density coils. A 3D SPGR sequence with intermittent fat suppression and Poisson-disc variable density k-space sampling was developed. 3 reconstructions included parallel imaging (ARC), compression sensing (L1-SPIRiT) and coil compressed (CC) L1-SPIRiT. CC-L1-SPIRiT showed better image quality performance for most qualitative assessments. Compressed sensing with fast image reconstruction is feasible in a pediatric clinical environment and can improve quality of structural delineation in pediatric MRI.

Abstract PDF

Keywords

abdominal acceleration acceptable adrenal adversely affect ages agreement alley alone analyses anatomical appearance approval arrow artery artifacts assess assessed assessment assessments back bile binomial black blinded blurred blurring bone boxes branch channel children clinical coil coils combination common compressed compression computer consecutive consent contrast cortex dashed delineation denoted density despite developed diagnostically diagram direct disc done duct electrical enabling engineering environment evaluation except fast feasible female gland good graded grader gray hardware hepatic iliac improve improved indicating individual individually instead inter intermittent least liver location look mark matrix middle minute minutes mostly observer overall paired pairs pancreatic parallel patient pediatric performance permits portal post qualitative quality radiologists radiology randomized rank rapid readout recently reconstruction reconstructions recruited recruitment reduces referred renal representative sample sampling scores scoring sensing short space spine spirit statistical statistically statistics structural structures subject superior suppression synthetic table thick transformed type validated variable vein veins versus virtual white years