Multi-modal pre-post treatment consistency learning for automatic segmentation and evaluation of the Circle of Willis

Zehang Lin, Yusheng Liu, Jiahua Wu, Da-Han Wang, Xu-Yao Zhang, Shunzhi Zhu

Abstract: Highlights•Propose CMSC-Net for consistent vascular segmentation across CTA and MRA.•Propose SC-ENet to quantify treatment efficacy by detecting vessel changes.•Introduce MPAM to generate spatially-aligned modality pairs.•Develop CMAM to enhance CTA feature extraction with MRA semantic guidance.•Design a novel loss for consistent vascular representation across modalities.