A medical image segmentation method and system based on a double-path U-shaped network

Abstract

The application provides a medical image segmentation method and system based on a double-path U-shaped network, relates to the technical field of image analysis, and comprises the following steps: acquiring first information, wherein the first information is a medical image set; constructing an encoder part, wherein the encoder part comprises a first path and a second path, the first path is an up-sampling path, and the second path is a down-sampling path; constructing a decoder part, wherein the decoder part is an up-sampling path; connecting the encoder part and the decoder part through a first convolution module to obtain a double-path U-shaped network, and a skip connection is arranged between the encoder part and the decoder part; inputting a training set and a test set obtained by preprocessing the first information into the double-path U-shaped network for training to obtain an optimal segmentation model. More low-dimensional features of an image are extracted through the up-sampling path added to the encoder part, and the low-dimensional features contain boundary structures and more fine structures of the image.

Images