Systematic investigation of Basic Data Augmentation Strategies on Histopathology Images

Abstract

Recent years have witnessed the rapid progress of deep neural networks. However, in supervised learning, the success of the models hinges on a large amount of training data. Therefore, data augmentation techniques were developed to increase the effective size of the training data. Using such techniques is especially important for domains where the amount of available data is limited. In digital pathology, data augmentation is therefore often applied to improve the performance of classifications. This work systematically investigates single data augmentation techniques on different datasets using multiple network architectures. Furthermore, it proposes guidelines on using data augmentation when training deep neural networks on histopathological data.