DiffuSAM: Diffusion-Based Prompt-Free SAM2 for Few-Shot and Source-Free Medical Image Segmentation

TL;DR

DiffuSAM introduces a diffusion-based method to adapt SAM2 for prompt-free medical image segmentation, enabling accurate, prompt-free segmentation across domains without extensive fine-tuning.

Contribution

It presents a novel diffusion prior that synthesizes segmentation embeddings compatible with SAM2, eliminating the need for prompts and improving domain transfer in medical imaging.

Findings

Achieves competitive performance on BTCV and CHAOS datasets.

Enables prompt-free segmentation with efficient training and inference.

Enforces spatial consistency across slices in volumetric data.

Abstract

Segmentation models such as Segment Anything Model (SAM) and SAM2 achieve strong prompt-driven zero-shot performance. However, their training on natural images limits domain transfer to medical data. Consequently, accurate segmentation typically requires extensive fine-tuning and expert-designed prompts. We propose DiffuSAM, a diffusion-based adaptation of SAM2 for prompt-free medical image segmentation. Our framework synthesizes SAM2-compatible segmentation mask-like embeddings via a lightweight diffusion-prior from off-the-shelf frozen SAM2 image features. The generated embeddings are integrated into SAM2's mask decoder to produce accurate segmentations, thereby eliminating the need for user prompts. The diffusion prior is further conditioned on previously segmented slices, enforcing spatial consistency across volumes. Evaluated on the BTCV and CHAOS datasets for CT and MRI under Source-Free Unsupervised Domain Adaptation (SF-UDA) and Few-Shot settings, DiffuSAM achieves competitive performance with efficient training and inference. Code is available upon request from the corresponding author.