Technical - Emerging CMR Methods - Acquisition & Reconstruction-Highly Accelerated Free-Breathing Real-time Phase Contrast Cardiovascular MRI via Complex-Difference Deep Learning

Highly Accelerated Free-Breathing Real-time Phase Contrast Cardiovascular MRI via Complex-Difference Deep Learning

Video Transcription

Video Summary

The study developed a deep learning framework to rapidly reconstruct highly accelerated radial real-time phase contrast MRI without breath-holding or ECG gating. Using training data from 510 patients and testing in 21 patients, the method produced better image quality than iterative GRASP reconstruction and greatly reduced reconstruction time. It also improved agreement with conventional ECG-gated segmented phase contrast for several hemodynamic measurements, though performance varied by metric. Overall, deep learning enabled fast, free-breathing real-time phase contrast MRI with promising clinical potential for hemodynamic assessment.

Keywords

deep learning

real-time phase contrast MRI

radial MRI reconstruction

hemodynamic assessment

free-breathing MRI