Brief Review of Medical Image Segmentation Based on Deep Learning

Abstract

In recent years, deep learning models based on Convolutional Neural Networks (CNN), such as U-Net, ResNet and VGG, have made outstanding achievements in the field of medical image segmentation, providing strong support for the diagnosis of related diseases.
However, most of the existing models assume that training data and test data meet the requirement of Independent and Identically Distributed (IID), while in practice, medical images from different imaging devices do not meet this requirement, that is, there is a problem of domain migration. As a result, the accuracy and stability of segmentation results are greatly reduced. Therefore, how to solve the problem of domain migration is of great significance to the practical application of the model. In this paper, the application of deep learning algorithm in medical image segmentation, feature extraction and classification is described. Secondly, the algorithm of deep learning processing multi-modal medical images is summarized. Finally, the problems existing in medical image diagnosis are pointed out, and the future development direction is prospected.