Lesion classification method based on organ-specific feature perception and regional feature enhancement

By introducing methods of organ-specific feature perception and regional feature enhancement in multi-organ lesion classification, the problem of insufficient model universality and accuracy in the prior art is solved, and a more efficient abdominal multi-organ lesion classification is achieved.

CN118736282BActive Publication Date: 2025-04-01DAYI ZHICHENG HIGH TECH CO LTD +1
2 Cites 0 Cited by

Patent Information

Application Number
CN202410764746.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-14
Publication Date
2025-04-01
Estimated Expiration
2044-06-14

AI Technical Summary

Technical Problem

In the multi-organ lesion classification, existing computer-assisted diagnostic methods cannot meet the complex and changeable task requirements in clinical practice due to the research limitations of a single organ or tissue, resulting in insufficient model universality and accuracy.

Method used

A multi-organ lesion classification method based on organ-specific feature perception and regional feature enhancement is proposed. The lesion characteristics of specific organs are extracted through the organ-specific feature perception module, and the lesion characteristics with significant distinction are retained through the regional feature enhancement module to construct a general multi-organ lesion classification model.

Benefits of technology

It improves the accuracy of the classification of multiple organ lesions in the abdomen, enhances the universality of the model, and reduces the occurrence of misdiagnosis and misdiagnosis.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention discloses a lesion classification method based on organ-specific feature perception and regional feature enhancement, which mainly solves the problem that the existing method will miss the lesion features with significant distinction in different organs. Its implementation scheme is: obtain an abdominal CT image data set and perform preprocessing of organ segmentation, normalization, and downsampling, and divide the preprocessed data into a training set and a test set; construct a multi-organ lesion classification model including an organ-specific feature perception module, a regional feature enhancement module, a multi-scale residual attention module, and a classifier; use a multi-organ lesion classification training set to iteratively train the lesion classification model; input the test set into the trained lesion classification model, and output the classification result. The present invention can extract the lesion features of different organs in a targeted manner, and maximize the retention of features with significant distinction, improve the accuracy of abdominal multi-organ lesion classification, and can be used for screening of lesions of multiple target organs in the human abdomen in CT images.
Need to check novelty before this filing date? Find Prior Art