TriAug: Out-of-Distribution Detection for Imbalanced Breast Lesion in Ultrasound
This addresses the challenge of detecting unseen breast lesion classes in clinical settings, which is an incremental improvement for medical imaging.
The paper tackled the problem of out-of-distribution detection for imbalanced breast lesion subtypes in ultrasound images, achieving an F1-score of 42.12% for in-distribution classification and an AUROC of 78.06% for OOD detection.
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%).