SAM-Driven Weakly Supervised Nodule Segmentation with Uncertainty-Aware Cross Teaching

TL;DR

This paper introduces a weakly supervised nodule segmentation framework leveraging SAM foundation models, adaptive pseudo-labeling, and uncertainty-aware cross-teaching to reduce annotation costs and improve segmentation accuracy in ultrasound images.

Contribution

It proposes a novel framework that uses aspect ratio annotations with SAM for pseudo-label generation and incorporates uncertainty-aware cross-teaching to enhance segmentation performance.

Findings

Outperforms existing methods on ultrasound datasets

Effective pseudo-label generation from minimal annotations

Robust segmentation with uncertainty-aware training

Abstract

Automated nodule segmentation is essential for computer-assisted diagnosis in ultrasound images. Nevertheless, most existing methods depend on precise pixel-level annotations by medical professionals, a process that is both costly and labor-intensive. Recently, segmentation foundation models like SAM have shown impressive generalizability on natural images, suggesting their potential as pseudo-labelers. However, accurate prompts remain crucial for their success in medical images. In this work, we devise a novel weakly supervised framework that effectively utilizes the segmentation foundation model to generate pseudo-labels from aspect ration annotations for automatic nodule segmentation. Specifically, we develop three types of bounding box prompts based on scalable shape priors, followed by an adaptive pseudo-label selection module to fully exploit the prediction capabilities of the foundation model for nodules. We also present a SAM-driven uncertainty-aware cross-teaching strategy. This approach integrates SAM-based uncertainty estimation and label-space perturbations into cross-teaching to mitigate the impact of pseudo-label inaccuracies on model training. Extensive experiments on two clinically collected ultrasound datasets demonstrate the superior performance of our proposed method.