MAE-SAM2: Mask Autoencoder-Enhanced SAM2 for Clinical Retinal Vascular Leakage Segmentation

We propose MAE-SAM2, a novel foundation model based framework for retinal vascular leakage segmentation in fluorescein angiography images. Due to the small size and dense distribution of the leakage areas, along with the limited availability of labeled clinical data, this segmentation task is significant challenge. To address these challenges, we integrate a self-supervised pre-training strategy based on Masked Autoencoders (MAE) with the foundation model MedSAM2. We further investigate several loss functions and identify a task-specific combined loss that yields the best performance. Extensive experiments and ablation studies demonstrate that MAE-SAM2 consistently outperforms multiple state-of-the-art methods, achieving the highest Dice score and Intersection-over-Union (IoU). These results highlight the potential of combining self-supervised pre-training with foundation models for challenging clinical image segmentation tasks.