Fig. 1: Visual comparison of segmentation results on representative IVUS images.

From left to right: original image (Image), ground-truth (GT), our model’s prediction (Ours), U-Net’s prediction, and UNETR's prediction. Our model’s segmentation (red overlay) aligns more closely with the ground-truth (blue overlay) across various plaque morphologies. Rows (A−E) show representative cases with different plaque morphologies.