CT image liver tumor area automatic segmentation method based on multi-branch network

Abstract

The invention discloses a CT image liver tumor area automatic segmentation method based on a multi-branch network. The method comprises the following steps: (1) establishing a training data set A containing an original CT image and a liver tumor manual segmentation result and direction information thereof; (2) constructing a deep convolutional multi-branch network model fusing a self-attention mechanism and direction information; (3) constructing a network loss function; (4) training the network by adopting the training data set A; and (5) segmenting a test image by using the trained network model to obtain a final liver tumor segmentation result. According to the full-automatic liver tumor segmentation method, a self-attention module and a direction correction module based on direction information are introduced into the convolutional network, so that the problems that the convolutional network is difficult to establish a long-distance target dependency relationship in liver tumor segmentation and tumor boundary recognition is inaccurate are solved; and the segmentation precision of the liver tumor is effectively improved.

Description

technical field [0001] The invention relates to the technical field of medical image processing, in particular to an automatic segmentation method for a CT image liver tumor region based on a multi-branch network. Background technique [0002] Liver tumor segmentation is an indispensable means to extract quantitative information (such as volume, shape, size, etc.) of liver diseased tissue in medical images, and plays a vital role in radiotherapy, surgical planning, and efficacy evaluation of liver tumors. Currently, the clinical segmentation of liver tumors is mainly done by physicians manually delineating the anatomical structures in the patient's CT image data. Since there are hundreds of CT slices for each patient, manual segmentation is time-consuming and labor-intensive, and is affected by the experience and knowledge level of the physicians. There are significant differences in the segmentation results of different physicians. Therefore, it is of great significance to...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the shortcomings and deficiencies of the prior art, the present invention integrates the self-attention module and the direction correction module based on direction information into the construction of the deep convolutional network, aiming to provide a multi-branch network-based automatic segmentation method for liver tumor regions in CT images, Solve the problem that the convolutional network is difficult to establish long-distance target dependence in the segmentation of liver tumors and the inaccurate identification of tumor boundaries, and improve the accuracy and efficiency of computer-aided diagnosis of liver diseases

Images