Examining the Capability of GANs to Replace Real Biomedical Images in Classification Models Training

TL;DR

This study investigates whether synthetic biomedical images generated by GANs can effectively replace real images in training classification models, showing minimal accuracy loss especially with deep learning methods.

Contribution

It demonstrates the feasibility of using GAN-generated images as substitutes for real data in training, with a detailed comparison across different models and image types.

Findings

Synthetic images cause less than 4% accuracy drop in deep learning models.

Traditional methods experience up to 13% accuracy decrease with synthetic data.

GANs can produce realistic biomedical images suitable for training classifiers.

Abstract

In this paper, we explore the possibility of generating artificial biomedical images that can be used as a substitute for real image datasets in applied machine learning tasks. We are focusing on generation of realistic chest X-ray images as well as on the lymph node histology images using the two recent GAN architectures including DCGAN and PGGAN. The possibility of the use of artificial images instead of real ones for training machine learning models was examined by benchmark classification tasks being solved using conventional and deep learning methods. In particular, a comparison was made by replacing real images with synthetic ones at the model training stage and comparing the prediction results with the ones obtained while training on the real image data. It was found that the drop of classification accuracy caused by such training data substitution ranged between 2.2% and 3.5% for deep learning models and between 5.5% and 13.25% for conventional methods such as LBP + Random Forests.