Linear Fusion-Based Dual-MR Contrast Enhancement for Improved Choroid Plexus Segmentation

Motivation: Combining T1w and T2w-FLAIR images via arithmetic operation-based fusion significantly enhanced image contrast, emphasizing the boundaries and improving tissue differentiation around the choroid plexus (ChP).
Goal(s): To improve segmentation accuracy of the ChP and evaluate the effectiveness and feasibility of the proposed technique across various DL models.
Approach: Dual-MR contrast fusion is conducted following image alignment, normalization, and linear computation between T1w and T2w-FLAIR images. Off-the-shelf DL models were used to perform ChP segmentation.
Results: Fused MR images achieved higher quantitative values (Hausdorff distances and Dice scores), with significant improvements in Relative Contrast and SNR compared to methods utilizing a single MR image.