Optimizing Respiratory Phase Binning for Improved Motion Correction in Abdominal Diffusion-Weighted MRI

Michelle Su, Cemre Ariyurek, Serge Vasylechko, Onur Afacan, Sila Kurugol

Abstract: Motion artifacts pose a significant challenge in abdominal diffusion-weighted magnetic resonance imaging (DW-MRI). Partitioning image slices into respiratory phase bins can reduce motion artifacts, but for DW-MRI, the standard binning method often leaves volumes lacking slices along the superior-inferior axis. As a result, longer scans are required for standard binning. In this study, we propose a new binning method to minimize missing slices without extending scans. We first optimize rigid partitions in the respiratory signal using dynamic programming and prefix sum approaches. Then, we create a probabilistic procedure to choose some slices to belong in multiple bins. Our method results in significantly fewer missing slices than standard binning (p<1.0× 10–15), yielding an average reduction of 81.74±7.58%. Compared to uncorrected images, our method reduces intra-subject variability in homogeneous regions (p<0.001) and improves lesion conspicuity. Thus, our method is effective in reducing missing slices, increasing anatomical accuracy, and shortening acquisition time.

External IDs:dblp:conf/isbi/SuAVAK25