Semi-supervised classification of radiology images with NoTeacher: A teacher that is not mean

Balagopal Unnikrishnan, Cuong Manh Nguyen, Shafa Balaram, Chao Li, Chuan Sheng Foo, Pavitra Krishnaswamy

Abstract: Highlights • Novel semi-supervised learning framework for 2D & 3D radiology image classification. • Adapted for class distribution mismatch between labelled and unlabelled training data. • Connections to theoretically principled multi-view learning and co-training approaches. • Realistic evaluations for relevance to practical clinical annotation workflows. • Superior semi-supervised performance for uni/multi-label tasks with X-Ray, CT, MRI. Abstract Deep learning models achieve strong performance for radiology image classification, but their practical application is bottlenecked by the need for large labeled training datasets. Semi-supervised learning (SSL) approaches leverage small labeled datasets alongside larger unlabeled datasets and offer potential for reducing labeling cost. In this work, we introduce NoTeacher, a novel consistency-based SSL framework which incorporates probabilistic graphical models. Unlike Mean Teacher which maintains a teacher network updated via a temporal ensemble, NoTeacher employs two independent networks, thereby eliminating the need for a teacher network. We demonstrate how NoTeacher can be customized to handle a range of challenges in radiology image classification. Specifically, we describe adaptations for scenarios with 2D and 3D inputs, with uni and multi-label classification, and with class distribution mismatch between labeled and unlabeled portions of the training data. In realistic empirical evaluations on three public benchmark datasets spanning the workhorse modalities of radiology (X-Ray, CT, MRI), we show that NoTeacher achieves over 90–95% of the fully supervised AUROC with less than 5–15% labeling budget. Further, NoTeacher outperforms established SSL methods with minimal hyperparameter tuning, and has implications as a principled and practical option for semi-supervised learning in radiology applications.