Synthetic image augmentation with generative adversarial network for enhanced performance in protein classifi cation

Proteins are complex macromolecules accountable for the biological processes in the cell. In biomedical research, the imagesof protein are extensively used in medicine. The rate at which these images are produced makes it diffi cult to evaluate themmanually and hence there exists a need to automate the system. The quality of images is still a major issue. In this paper, wepresent the use of diff erent image enhancement techniques that improves the contrast of these images. Besides the qualityof images, the challenge of gathering such datasets in the fi eld of medicine persists. We use generative adversarial networksfor generating synthetic samples to ameliorate the results of CNN. The performance of the synthetic data augmentationwas compared with the classic data augmentation on the classifi cation task, an increase of 2.7% in Macro F1 and 2.64%in Micro F1 score was observed. Our best results were obtained by the pretrained Inception V4 model that gave a fi vefoldcross-validated macro F1 of 0.603. The achieved results are contrasted with the existing work and comparisons show thatthe proposed method outperformed. KCI Citation Count: 0