Multi-scale DC-CUNets liver tumor segmentation method based on bottleneck structure

Abstract

The invention discloses a multi-scale DC-CUNets liver tumor segmentation method based on a bottleneck structure, and aims to solve the technical problem of insufficient liver tumor segmentation precision in the prior art. The method comprises the following steps: obtaining a liver mask from a vein CT image by utilizing U-Net based on a bottleneck structure; performing mask operation on the CT images in the arterial phase and the venous phase and the liver mask to obtain liver regions of interest in the arterial phase and the venous phase; processing liver regions of interest in the arterial phase and the venous phase by using dual-channel cascade U-Nets to obtain deep image features in the arterial phase and the venous phase; and performing feature fusion on the deep image features in the arterial phase and the venous phase, processing the fused feature blocks by using a softmax layer, and outputting a liver tumor segmentation probability graph. According to the method, liver tumor segmentation can be rapidly and accurately carried out.

Description

technical field [0001] The invention relates to a multi-scale DC-CUNets liver tumor segmentation method based on a bottleneck structure, and belongs to the technical field of medical image processing. Background technique [0002] As a common malignant tumor disease, liver cancer is one of the most common cancers in the world. Accurate diagnosis of liver tumors based on CT effects has very important guiding significance for liver cancer research, clinical diagnosis and surgical treatment. At present, more and more studies have used computer image processing technology as an auxiliary diagnosis and treatment method to segment liver and tumors from CT images. However, information such as liver size and location varies from person to person, and intraperitoneal Due to the complex connection of organs and tissues in medical imaging, the low contrast of medical images, the diversity and spread of tumors, the blurred boundaries of liver and tumors, and the uneven density distribut...

AI Technical Summary

Problems solved by technology

[0003] In order to solve the problem of insufficient segmentation accuracy of liver tumors in the prior art, the present invention proposes a multi-scale DC-CUNets liver tumor segmentation method based on the bottleneck structure, and uses the multi-scale atrous convolution kernel depth bottleneck structure to transform the dual-channel level Combined U-Nets to speed up network convergence while improving liver tumor segmentation accuracy

Images