ProtoSAM: One-Shot Medical Image Segmentation With Foundational Models

TL;DR

ProtoSAM introduces a novel one-shot medical image segmentation framework that combines prototypical networks with SAM, achieving state-of-the-art results without fine-tuning on various datasets.

Contribution

It presents a new one-shot segmentation method that integrates prototypical networks with SAM, enabling effective medical image segmentation from a single example without additional training.

Findings

Achieves state-of-the-art performance on multiple datasets.

Operates effectively with only one image example.

Does not require fine-tuning of the foundation model.

Abstract

This work introduces a new framework, ProtoSAM, for one-shot medical image segmentation. It combines the use of prototypical networks, known for few-shot segmentation, with SAM - a natural image foundation model. The method proposed creates an initial coarse segmentation mask using the ALPnet prototypical network, augmented with a DINOv2 encoder. Following the extraction of an initial mask, prompts are extracted, such as points and bounding boxes, which are then input into the Segment Anything Model (SAM). State-of-the-art results are shown on several medical image datasets and demonstrate automated segmentation capabilities using a single image example (one shot) with no need for fine-tuning of the foundation model. Our code is available at: https://github.com/levayz/ProtoSAM