Deep Learning for Medical Image Diagnosis: A Review of Methods, Applications, and Challenges

Medical image diagnosis is an important way of early detection of diseases. However, classical manual image reading is time-consuming and subject to human factor. The introduction of Deep Learning (DL) tools, and especially Convolutional Neural Networks (CNNs), drastically changed automated analysis in medical images. This article surveys the major deep learning concepts for medical image analysis, and provides an overview of current and potential future directions. DL models have demonstrated tremendous success in disease diagnosis with X- rays, MRI/CT and histopathological images leading to faster and accurate results. The paper also discusses challenges in terms of data privacy, interpretability and generalization.