Efficient Active Learning for Image Classification and Segmentation using a Sample Selection and Conditional Generative Adversarial Network

TL;DR

This paper introduces an active learning framework that combines sample selection and conditional GANs to efficiently train deep learning models for medical image classification and segmentation, reducing data requirements significantly.

Contribution

It presents a novel active learning approach utilizing cGANs and Bayesian neural networks to select informative samples, improving efficiency in medical image analysis tasks.

Findings

Achieves state-of-the-art performance using only 35% of data

Reduces training data needs and annotation effort

Demonstrates effectiveness on chest X-ray image classification

Abstract

Training robust deep learning (DL) systems for medical image classification or segmentation is challenging due to limited images covering different disease types and severity. We propose an active learning (AL) framework to select most informative samples and add to the training data. We use conditional generative adversarial networks (cGANs) to generate realistic chest xray images with different disease characteristics by conditioning its generation on a real image sample. Informative samples to add to the training set are identified using a Bayesian neural network. Experiments show our proposed AL framework is able to achieve state of the art performance by using about 35% of the full dataset, thus saving significant time and effort over conventional methods.