Medical Image Classification on Imbalanced Data Using ProGAN and SMA-Optimized ResNet: Application to COVID-19

TL;DR

This work tackles the challenge of imbalanced medical imaging data for COVID-19 detection in chest X-rays by generating per-class synthetic images with ProGAN and integrating them with real data using a weighted injection strategy. An SMA-based hyper-parameter optimization framework tunes a pre-trained ResNet50V2 classifier to perform robust four-class and binary classifications under imbalance. The approach yields improved cross-validated accuracy (approximately 95% for four-class and 98% for two-class scenarios) and demonstrates a practical data-augmentation pathway during pandemics. Generating about 40K synthetic images across classes and optimizing injection ratios show that carefully balanced synthetic data can meaningfully boost diagnostic performance in imbalanced medical imaging tasks.

Abstract

The challenge of imbalanced data is prominent in medical image classification. This challenge arises when there is a significant disparity in the number of images belonging to a particular class, such as the presence or absence of a specific disease, as compared to the number of images belonging to other classes. This issue is especially notable during pandemics, which may result in an even more significant imbalance in the dataset. Researchers have employed various approaches in recent years to detect COVID-19 infected individuals accurately and quickly, with artificial intelligence and machine learning algorithms at the forefront. However, the lack of sufficient and balanced data remains a significant obstacle to these methods. This study addresses the challenge by proposing a progressive generative adversarial network to generate synthetic data to supplement the real ones. The proposed method suggests a weighted approach to combine synthetic data with real ones before inputting it into a deep network classifier. A multi-objective meta-heuristic population-based optimization algorithm is employed to optimize the hyper-parameters of the classifier. The proposed model exhibits superior cross-validated metrics compared to existing methods when applied to a large and imbalanced chest X-ray image dataset of COVID-19. The proposed model achieves 95.5% and 98.5% accuracy for 4-class and 2-class imbalanced classification problems, respectively. The successful experimental outcomes demonstrate the effectiveness of the proposed model in classifying medical images using imbalanced data during pandemics.

Figures (12)

• Figure 1: Basic diagram of the performed methodology.
• Figure 2: General structure of generator output size and critic input size networks throughout training procedure.
• Figure 3: Critic & Generator losses versus iteration at stage #1 for Covid-19 class
• Figure 4: Critic & Generator losses versus iteration at stage #2 for Covid-19 class
• Figure 5: Critic & Generator losses versus iteration at stage #3 for Covid-19 class