How to best match voxels – evaluating different sequential co-registration strategies for ultra-high b-value DWI in multiparametric breast MRI

Motivation: Breast MRI increasingly includes ultra-high b-value DWI, but technical and patient-related challenges can cause mismatches to morphologic sequences. Co-registration strategies are needed to reliably apply AI-algorithms to the data. Goal(s): The study examines how different co-registration orders for ultra-high b-value acquisitions in multiparametric breast MRI impact co-localization to morphologic sequences, both quantitatively and qualitatively. Approach: This IRB-approved retrospective study of 144 multiparametric breast MRI exams with ultra-high b-value DWI, assessed different co-registration algorithms using ANTs library methods quantitatively and qualitatively. Results: The sequential arrangement of contrasts used during co-registration significantly affects cross-sequence congruence of lesions in multiparametric breast MRI incorporating ultra-high b-value DWI. Impact: The study highlights the relevance of considering different sequential arrangements used during co-registration in ultra-high b-value DWI.