Reconstruction of High-Resolution, Quantitative MRI Images in a Clinical Setting

Curtis Johnson, University of Illinois at Urbana-Champaign

Usage Details

co-PIs: Graham, R. Huesmann, Carle Foundation Hospital; Bradley P. Sutton, University of Illinois at Urbana–Champaign; Matthew D. J. McGarry, Dartmouth College

High-resolution, quantitative MRI methods can provide critical information for treatment planning in medicine, such as in pre-surgical examination of epilepsy. However, the image reconstruction procedures associated with these methods are too computationally extensive for the scanner console hardware. This results in reconstruction times on the order of days, limiting the use in a clinical setting. The Blue Waters system is uniquely suited to aid in overcoming this roadblock by providing access to a massively parallel architecture than can be harnessed to achieve “real-time” reconstructions that fit in the clinical workflow. This project will implement codes on Blue Waters and benchmark their performance with the goal of reconstruction in 1.5 hours, and will then use this accelerated processing for reconstructing datasets acquired during pre-surgical MRI examinations of patients with mesial temporal lobe epilepsy. Ultimately, Blue Waters will allow for improved treatment of patients by enabling high-resolution, quantitative MRI acquisitions in a clinical setting.