MRI image segmentation by fully convolutional networks

Abstract

With the development of various imaging technologies, medical imaging has been playing more important roles on providing scientific proof for doctors to make decisions on clinical diagnosis. At the same time, it is very important to excavate valuable information hidden in those images and take over some auxiliary medical works from doctors by the computer. Therefore, a large number of image segmentation methods, including some classic algorithms have been proposed and some of them perform well. In this paper, we built a deep convolutional neural network to segment the MRI brain images. Results show that the network has a good performance on segmentation of the gray and white matter of brains, it also had a good generalization ability.