CT Image Segmentation System Based on Attentional Convolutional Neural Network

Abstract

The present invention provides a CT image segmentation system based on attentional convolutional neural network, including using parallel convolutional neural network to gradually reduce the size of the feature map of the input image, multiplexing the network layer and intercepting and merging the features of each layer, Realize the feature encoding module that simultaneously extracts image semantic information and spatial information; use pooling to generate attention features, and further refine and refine the semantic information features extracted by the feature encoding module; use maximum pooling and averaging Pooling in parallel, which fuses the refined semantic information features with the semantic information and spatial information features spliced ​​by the feature encoding module to form a feature fusion pooling attention module of the attention feature map; use the convolution module and the upsampling module , a feature map decoding module that gradually and finely restores the attention feature map to the size of the input image. The present invention realizes efficient and accurate image segmentation by fusing attention modules.

Description

technical field [0001] The invention relates to the technical field of image understanding, in particular to a CT image segmentation system based on an attention convolutional neural network. Background technique [0002] Image segmentation is an important basic research problem in the field of computer vision, and medical image segmentation, as an application of image segmentation, can accurately locate the lesions of a large number of patients in a short time and quickly. Therefore, how to effectively apply image segmentation technology to medical images has become the main task of researchers. [0003] Medical image segmentation classifies the semantic expressions in the image pixel by pixel by extracting medical image features. Medical image segmentation needs to accurately locate the object, the class it belongs to, and the location of the object, and it needs to clearly demarcate the object boundary to distinguish different types of objects. [0004] At present, ther...

AI Technical Summary

Problems solved by technology

However, there are some defects in the pyramid structure. For example, for the extracted information, what is the part that really needs to be paid attention to by the network, which is not well explained by the pyramid structure.

Images