USL-Net: Uncertainty Self-Learning Network for Unsupervised Skin Lesion Segmentation

TL;DR

The paper tackles unsupervised skin lesion segmentation, addressing artifacts like hair noise and subtle edge differences that hinder label-free learning. It introduces USL-Net, which first extracts features via contrastive learning and constructs Class Activation Maps to derive saliency-based pseudo-labels, while treating uncertain intermediate regions as unlabeled. It further refines foreground pseudo-labels through connectivity and centrality detection and enhances performance with cycle refining. Experimental evaluation on ISIC-2017, ISIC-2018, and PH2 demonstrates competitive results with weakly supervised and supervised methods and superiority over other unsupervised approaches.

Abstract

Unsupervised skin lesion segmentation offers several benefits, including conserving expert human resources, reducing discrepancies due to subjective human labeling, and adapting to novel environments. However, segmenting dermoscopic images without manual labeling guidance presents significant challenges due to dermoscopic image artifacts such as hair noise, blister noise, and subtle edge differences. To address these challenges, we introduce an innovative Uncertainty Self-Learning Network (USL-Net) designed for skin lesion segmentation. The USL-Net can effectively segment a range of lesions, eliminating the need for manual labeling guidance. Initially, features are extracted using contrastive learning, followed by the generation of Class Activation Maps (CAMs) as saliency maps using these features. The different CAM locations correspond to the importance of the lesion region based on their saliency. High-saliency regions in the map serve as pseudo-labels for lesion regions while low-saliency regions represent the background. However, intermediate regions can be hard to classify, often due to their proximity to lesion edges or interference from hair or blisters. Rather than risk potential pseudo-labeling errors or learning confusion by forcefully classifying these regions, we consider them as uncertainty regions, exempting them from pseudo-labeling and allowing the network to self-learn. Further, we employ connectivity detection and centrality detection to refine foreground pseudo-labels and reduce noise-induced errors. The application of cycle refining enhances performance further. Our method underwent thorough experimental validation on the ISIC-2017, ISIC-2018, and PH2 datasets, demonstrating that its performance is on par with weakly supervised and supervised methods, and exceeds that of other existing unsupervised methods.