TriAug: Out-of-Distribution Detection for Imbalanced Breast Lesion in Ultrasound

TL;DR

This work addresses out-of-distribution detection for imbalanced breast lesion subtypes in ultrasound imaging. It introduces TriAug, a triplet-state augmentation strategy, combined with a Balanced Sphere Loss to learn discriminative embeddings under long-tailed class distributions and robust OOD rejection. On an in-house breast US dataset, the approach achieves a notable improvement in both ID classification (F1 ≈ 42.1%) and OOD detection (AUROC ≈ 78.1%), outperforming state-of-the-art baselines. The results suggest that TriAug plus the sphere-based regularization enhances reliability of computer-aided diagnosis in medical imaging by flagging OOD subtypes for human review and mitigating dataset imbalance.

Abstract

Different diseases, such as histological subtypes of breast lesions, have severely varying incidence rates. Even trained with substantial amount of in-distribution (ID) data, models often encounter out-of-distribution (OOD) samples belonging to unseen classes in clinical reality. To address this, we propose a novel framework built upon a long-tailed OOD detection task for breast ultrasound images. It is equipped with a triplet state augmentation (TriAug) which improves ID classification accuracy while maintaining a promising OOD detection performance. Meanwhile, we designed a balanced sphere loss to handle the class imbalanced problem. Experimental results show that the model outperforms state-of-art OOD approaches both in ID classification (F1-score=42.12%) and OOD detection (AUROC=78.06%).

Figures (2)

• Figure 1: ID samples vs OOD samples of breast lesions in US.
• Figure 2: Schematic of the proposed framework.