Information Gain Sampling for Active Learning in Medical Image Classification

TL;DR

This paper introduces an information-theoretic active learning method, AEIG, that efficiently selects medical images for labeling, significantly reducing annotation efforts while maintaining high classification performance.

Contribution

The paper proposes AEIG, an adapted expected information gain approach that accounts for class imbalance, improving active learning efficiency in medical image classification.

Findings

AEIG outperforms popular baselines like CoreSet and maximum entropy sampling.

AEIG achieves ~95% of full data performance with only 19% of training data.

The method can be integrated into existing deep learning classifiers.

Abstract

Large, annotated datasets are not widely available in medical image analysis due to the prohibitive time, costs, and challenges associated with labelling large datasets. Unlabelled datasets are easier to obtain, and in many contexts, it would be feasible for an expert to provide labels for a small subset of images. This work presents an information-theoretic active learning framework that guides the optimal selection of images from the unlabelled pool to be labeled based on maximizing the expected information gain (EIG) on an evaluation dataset. Experiments are performed on two different medical image classification datasets: multi-class diabetic retinopathy disease scale classification and multi-class skin lesion classification. Results indicate that by adapting EIG to account for class-imbalances, our proposed Adapted Expected Information Gain (AEIG) outperforms several popular baselines including the diversity based CoreSet and uncertainty based maximum entropy sampling. Specifically, AEIG achieves ~95% of overall performance with only 19% of the training data, while other active learning approaches require around 25%. We show that, by careful design choices, our model can be integrated into existing deep learning classifiers.