Medical Image Segmentation Method Based on Residual Fully Convolutional Neural Network Based on Attention Mechanism

Abstract

The invention provides a medical image segmentation method based on an attention mechanism residual type full convolution neural network, which preprocesses the medical image to be segmented; constructs a residual type full convolution neural network based on an attention mechanism, including features Graph contraction network, attention network, feature map expansion network group; input the training set data into the residual full convolutional neural network for training to obtain the learned convolutional neural network model; input the test set data into the learned convolutional neural network The neural network model performs image segmentation to obtain the segmented image; this method uses the attention network to effectively transfer the image features extracted from the feature map shrinking network to the feature map expansion network, which solves the problem of lack of image deconvolution in the process of image deconvolution. At the same time, the attention network can also suppress the image areas that are not related to the segmentation target in the low-level feature map, reducing the redundancy of the image and increasing the accuracy of image segmentation.

Description

technical field [0001] The invention relates to a medical image segmentation method based on an attention mechanism and a residual type full convolution neural network. Background technique [0002] Medical image segmentation is a key issue to determine whether medical images can provide a reliable basis in clinical diagnosis and treatment. The development of medical image segmentation technology not only affects the development of other related technologies in medical image processing, such as visualization, 3D reconstruction, etc., but also occupies an extremely important position in the analysis of biomedical images. In recent years, medical image segmentation technology has made significant progress due to the application of deep learning algorithms in medical image segmentation. Medical image segmentation is generally modeled as a pixel-level multi-classification problem, where the goal is to classify each pixel of an image into one of a number of predefined classes. ...

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is to provide a medical image segmentation method based on a residual type fully convolutional neural network based on an attention mechanism, which solves the problem in the prior art that the spatial detail information is lost during the image deconvolution process, resulting in a lack of Spatial features of images, which simultaneously lead to the problem of excessive and redundant usage of computing resources and model parameters

Images