Self-Training with Improved Regularization for Sample-Efficient Chest X-Ray Classification
This addresses the need for sample-efficient automated diagnostic assistants in healthcare, though it appears incremental by combining existing techniques like data augmentation and self-training.
The paper tackled the problem of building accurate AI models for chest X-ray classification with limited labeled data, severe class imbalances, and multi-disease prediction, achieving performance matching classifiers trained on large-scale data using 85% less labeled data.
Automated diagnostic assistants in healthcare necessitate accurate AI models that can be trained with limited labeled data, can cope with severe class imbalances and can support simultaneous prediction of multiple disease conditions. To this end, we present a deep learning framework that utilizes a number of key components to enable robust modeling in such challenging scenarios. Using an important use-case in chest X-ray classification, we provide several key insights on the effective use of data augmentation, self-training via distillation and confidence tempering for small data learning in medical imaging. Our results show that using 85% lesser labeled data, we can build predictive models that match the performance of classifiers trained in a large-scale data setting.