Application of Machine Learning-based Iterative Image Reconstruction Algorithms for MRI Scan Time Reduction in Brain and Spine

published on August 9, 2018

Collaborative work with Lawrence N. Tanenbaum MD FACR, RadNet

PURPOSE

This study aims to evaluate the scan time shortening potential of iQMR, a novel 3D image enhancement algorithm, in brain, lumbar spine, and cervical spine MRI.

iQMR provides image enhancement and increased Signal-to-Noise Ratio (SNR), thus permitting substantial scan time reduction, without changing the imaging contrast, resolution, or scanner hardware or software.

This study reports findings of 73 patients (37 brain, 19 lumbar spine, 17 cervical spine) who underwent MRI scans on 4 different scanners at 3 sites. Image quality, diagnostic quality, artifacts’ appearance, and brain grey-white (GW) matter differentiation of iQMR-processed images acquired by short MRI protocols were compared to images acquired by the site’s standard MRI protocols.