Medical image segmentation method based on dual-path U-shaped convolutional neural network

Abstract

The present invention discloses a medical image segmentation method based on a dual-path U-shaped convolutional neural network. A dual-path module and a U-shaped network are combined, original features are reused through stacking operation, and features are added to generate new features so as to improve the network training precision; the U-shaped convolutional neural network configuration is employed to perform transverse feature stacking between blocks so as to greatly improve the network training convergence speed and more rapidly obtain an ideal training model. The medical image segmentation method can meet the demands of data size of medical images requiring being processed is large and high processing efficiency and high precision requirements and can achieve a very good performancein the biomedicine detection application.

Description

technical field [0001] The invention belongs to the technical field of medical image segmentation, and in particular relates to the design of a medical image segmentation method based on a dual-channel U-shaped convolutional neural network. Background technique [0002] Medical image segmentation includes manual segmentation, semi-automatic segmentation and fully automatic segmentation. Manual segmentation is difficult and the process is cumbersome. Although the accuracy is high, it largely depends on the experience and knowledge of the operator, and the segmentation results are difficult to reproduce; semi-automatic segmentation It is an interactive method that combines manual and computer processing, which allows human interactive operation to provide some useful information, and then the computer performs segmentation and processing. Semi-automatic segmentation also includes methods such as Gragh Cut (graph cut), CRF (conditional random field) and level set (level set). T...

AI Technical Summary

Problems solved by technology

[0003] Traditional image segmentation technology, whether it is manual segmentation or semi-automatic segmentation, requires manual input, the process is cumbersome, and largely depends on the experience and knowledge of the operator, and the segmentation results are difficult to reproduce, and fully automatic segmentation cannot be achieved. efficiency and precision

However, the existing fully automatic segmentation methods can only be limited to the 2D level, and the existing 3D network structure needs to be optimized.

Therefore, so far, there is no objective standard for accurately evaluating the success of segmentation.

Images