Medical Image Synthetic Data Augmentation Using GAN
Huijuan Zhang, Zongrun Huang, Zhongwei Lv
Abstract
Most medical image datasets are limited and unbalanced. The classification method based on deep neural networks is prone to under-fitting or over-fitting problems on such data sets, which affects the final classification performance. This paper proposes a synthetic data augmentation method based on progressive generative adversarial network, which aims to solve the problem that deep neural networks are difficult to train on small-scale medical image datasets. Experimental results show that the data synthesized by this method is better than existing methods. The classification performance using only classic data augmentation yielded 76.8% sensitivity and 88.4% specificity. By applying the synthetic data augmentation, the results significantly increased to 84.2% sensitivity and 92.1% specificity.