Evaluations of Deep Learning Methods for Pathology Image Classification

While the state-of-the-art deep neural networks have been shown their effectiveness to achieve the general image classification tasks, the performance evaluations of these networks on the pathology image classification have not been well discussed. In this paper, we aim to evaluate the state-of-the-art deep learning methods including the convolutional neural network, deep ensemble network and transformers on several public pathology datasets. By comparing these methods, we empirically find that the transformers and deep ensemble network serve as good backbone networks for pathology image classification.