Are foundation models efficient for medical image segmentation?

TL;DR

This study compares the efficiency of foundation models like SAM with self-supervised learning methods for cardiac ultrasound segmentation, finding that SAM is less efficient and performs poorly in this medical imaging task.

Contribution

The paper provides a direct comparison between foundation models and SSL methods in medical image segmentation, highlighting the limitations of foundation models in this domain.

Findings

SAM performed poorly compared to SSL

SAM required more labeling and compute resources

SSL was more efficient for cardiac ultrasound segmentation

Abstract

Foundation models are experiencing a surge in popularity. The Segment Anything model (SAM) asserts an ability to segment a wide spectrum of objects but required supervised training at unprecedented scale. We compared SAM's performance (against clinical ground truth) and resources (labeling time, compute) to a modality-specific, label-free self-supervised learning (SSL) method on 25 measurements for 100 cardiac ultrasounds. SAM performed poorly and required significantly more labeling and computing resources, demonstrating worse efficiency than SSL.