Esophageal varicosity classification system based on LightGBM and feature fusion

Abstract

The invention discloses an esophageal varicosity classification system based on LightGBM and feature fusion. The esophageal varicosity classification system comprises a segmentation module for segmenting and extracting regions of interest of the liver, the spleen and the esophagus in a CT image; a feature extraction module performing radiomics feature extraction on the region-of-interest image ofeach part; a first weight distribution module distributing equal weights to the radiomics characteristics of each part to obtain a first characteristic matrix; a second weight distribution module judging the importance of each part to the varicose veins of the affected esophagus by adopting a LightGBM method according to the radiomics characteristics of each part, and performs weighted fusion on the radiomics characteristics of each part according to the importance to obtain a second characteristic matrix; a classification module training a LightGBM classification model for the first feature matrix and the second feature matrix, and classifies whether the to-be-detected CT image suffers from esophageal varicose veins. An esophageal varicosity classification model based on LightGBM and feature fusion is constructed on the basis of radiomics, the importance of each part is brought into play, and the classification performance is improved.

Description

technical field [0001] The invention relates to the technical field of medical image classification, in particular to an esophageal varices classification system based on LightGBM and feature fusion. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] Medical image classification provides an important basis for clinical judgment. With the development of medical image technology, how to combine medical images with machine learning to efficiently achieve correct image classification has become a research hotspot in the field of medicine and machine learning. In the current field of medical image classification, there are few studies on the risk classification of esophageal varices. [0004] Esophageal varices are collaterals of the portosystemic venous system, connecting the portosystemic circulation. Esophageal varices are the result of portal...

AI Technical Summary

Problems solved by technology

In summary, the inventors believe that CT is the most commonly used technique to detect liver cirrhosis and its complications, and radiomics has been applied in many medical image classification studies. There are still few researches on non-invasive diagnosis

Images