Gastrointestinal medical image multi-classification method based on variance filtering algorithm

Abstract

The invention discloses a gastrointestinal medical image multi-classification method based on a variance filtering algorithm. The gastrointestinal medical image multi-classification method comprises the following steps of 1) firstly, performing feature extraction on an input gastroscope picture and an input enteroscope picture, 2) carrying out normalization and dimension reduction processing on the extracted features, 3) fusing the features in pairs to obtain six groups of new features, 4) performing 10-fold cross validation on each group of features by adopting a LinearSVC multi-classification algorithm to obtain the maximum accuracy and four corresponding weights of each feature, 5) putting the weights of the features together to calculate a variance, and then sorting the features according to the variance, 6) carrying out IFS-LinearSVC multi-classification on the sorted features to obtain an accuracy rate acc and used features, and 7) selecting two feature fusion methods with relatively high accuracy as classification models, and carrying out multi-classification on the gastrointestinal endoscope medical image. According to the method, iconography evaluation can be carried out more accurately, the repeatability is high, the working efficiency can be remarkably improved, and misdiagnosis and missed diagnosis are reduced.

Description

technical field [0001] The invention belongs to the technical field of medical image classification, in particular to a multi-classification method for gastrointestinal medical images based on a variance filtering algorithm. Background technique [0002] Artificial intelligence is one of the key technologies for processing medical images in the world. Although endoscopy programs have reduced mortality from gastrointestinal malignancies, they remain the leading cause of death worldwide and remain a global economic burden. Improving the detection rate of GI tumors, optimizing treatment strategies, high-quality endoscopy to identify GI tumors and classifying the relationship between benign and malignant lesions are essential for gastroenterologists. Especially inexperienced endoscopists may miss gastrointestinal lesions, which may have serious consequences for patients. Artificial intelligence technology can assist doctors to detect subtle changes in endoscopic images, and ca...

AI Technical Summary

Problems solved by technology

Especially inexperienced endoscopists may miss gastrointestinal lesions, which may have serious consequences for patients

Images