Unsupervised Approaches for Out-Of-Distribution Dermoscopic Lesion Detection

TL;DR

This paper introduces an unsupervised out-of-distribution detection method, SimCLR-LOF, for dermoscopic lesion images, demonstrating competitive performance against supervised approaches on the ISIC 2019 dataset.

Contribution

It presents a novel unsupervised OOD detection algorithm, SimCLR-LOF, combining semantic feature learning with LOF scoring for medical image analysis.

Findings

SimCLR-LOF performs competitively with state-of-the-art supervised methods.

Unsupervised OOD detection is effective on complex medical data.

Preliminary results show promise for unsupervised approaches in medical diagnostics.

Abstract

There are limited works showing the efficacy of unsupervised Out-of-Distribution (OOD) methods on complex medical data. Here, we present preliminary findings of our unsupervised OOD detection algorithm, SimCLR-LOF, as well as a recent state of the art approach (SSD), applied on medical images. SimCLR-LOF learns semantically meaningful features using SimCLR and uses LOF for scoring if a test sample is OOD. We evaluated on the multi-source International Skin Imaging Collaboration (ISIC) 2019 dataset, and show results that are competitive with SSD as well as with recent supervised approaches applied on the same data.