Towards real-time bulk-motion detection in cardiac MRI

Motivation: If undetected, bulk-motion during free-running cardiac MR image acquisition decreases image quality. Goal(s): The goal of this work is to test the feasibility of a real-time bulk-motion detection framework. Approach: The framework uses a PCA-based model and data consistency measures. It was evaluated in a retrospective volunteer study of free-running cardiac MR data and instructed bulk-motion events. Results: The results show that bulk-motion was detected correctly within a processing-time of 6ms per shot, thus advancing the hypothesis that prospective bulk-motion detection can be accomplished for cardiac free-running MRI when the framework is implemented inline. Impact: When implemented inline on the MRI scanner, the proposed bulk-motion detection framework can prevent inefficient workflows and compromised diagnostic image quality. Real-time bulk-motion detection paves the way for an automated prospective adaptation of MRI sequence parameters on the fly.