Improving Breast Cancer Diagnosis by Incorporating Transcytolemmal Water Exchange in Time-dependent Diffusion MRI

Motivation: Microstructural imaging based on IMPULSED is effective in predicting breast cancer molecular subtypes. However, IMPULSED ignored transcytolemmal water exchange. Goal(s): To evaluate the effect of water exchange on the identification of breast cancer molecular subtypes. Approach: Time-dependent diffusion MRI was conducted on breast cancer patients. Microstructural parameters were extracted from the dMRI signals by IMPULSED, JOINT and EXCHANGE, respectively. The performance of molecular subtyping was compared across the three models. Results: Incorporating water exchange into dMRI signal analysis improved the performance in predicting PR, HER2 and Ki67 status, as well as identifying HER2-enriched subtype. Impact: Our findings proved that transcytolemmal water exchange impact breast cancer diagnosis, highlighting the importance of model selection of time-dependent diffusion MRI tailored to a specific clinical task.