Semi-supervised learning-based portal vein detection and positioning method and system

Abstract

The invention provides a semi-supervised learning-based portal vein detection positioning method and system, and the method comprises the steps: carrying out model training through employing a semi-supervised learning model, carrying out feature extraction of unlabeled data, and taking the probability that whether a feature is a portal vein or not as a soft label; carrying out model training through a small number of label samples and a large number of label-free sample data, so that the manual labeling cost is greatly reduced; in addition, focus attention is formed through series connection of channel attention and space attention according to the position feature of the portal vein structure, so that priori knowledge is input into the convolutional neural network to guide model training, and the defects of existing portal vein detection are overcome; a loss function is replaced, a local loss function is selected as the loss function of the classifier, and normal training can still be carried out under the condition that positive and negative samples are unbalanced.

Description

technical field [0001] The present invention relates to the technical field of medical image processing, in particular to a method and system for detecting and locating a hepatic portal vein based on semi-supervised learning. Background technique [0002] As a very important vein in the human body, the hepatic portal vein provides very important support for the normal life activities of the human body. In clinical medicine, it is often necessary to detect and locate the hepatic portal vein in CT images. [0003] Previously, the detection and positioning of the hepatic portal vein in CT images was usually manually identified by professional physicians. While the work efficiency was low, the accuracy completely depended on the physician's work experience, and there was an inevitable problem of positioning errors. Nowadays, with the development of deep learning technology, it is the general trend to apply it to target detection and positioning of CT images. There are many type...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a method and system for hepatic portal vein detection and positioning based on semi-supervised learning, which aims to solve the problems of long manual labeling and poor pertinence of hepatic portal vein detection and positioning in the prior art, and reduce the cost of manual labeling , and targeted to improve positioning accuracy

Images