A multi-gene mutation prediction method based on multi-task and multi-instance learning

By combining multi-task and multi-instance learning with deep learning networks such as transformer and MobileNet, the problems of accuracy and resource consumption in multi-gene mutation detection in pathological image data are solved, achieving efficient and accurate multi-gene mutation prediction, especially showing excellent performance in TP53 gene mutation detection.

CN115719615BActive Publication Date: 2026-05-26NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
2 Cites 0 Cited by

Patent Information

Application Number
CN202211403709.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-11-10
Publication Date
2026-05-26
Estimated Expiration
2042-11-10

AI Technical Summary

Technical Problem

Existing technologies struggle to efficiently utilize pathological image data for simultaneous multi-gene mutation detection, particularly for the accurate diagnosis of KRAS and TP53 gene mutations in colon cancer. Furthermore, traditional methods consume significant computational resources and lack sufficient accuracy.

Method used

A combined approach of multi-task and multi-instance learning is adopted, combining the transformer and MobileNet modules. The feature matrix of pathological images is constructed through two-layer multi-instance learning, and the deep learning network is trained using the cross-entropy loss function and the Adam optimizer to achieve the prediction of multi-gene mutations.

Benefits of technology

It improves the accuracy and efficiency of multi-gene mutation detection and saves computational resources, especially showing a significant improvement in TP53 gene mutation detection, with significantly improved recall and accuracy.

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

Abstract

This invention discloses a multi-gene mutation prediction method based on a combination of multi-task and multi-instance learning. Belonging to the fields of digital image analysis, pathology, and machine learning, the specific steps are as follows: Preprocessing existing pathological image data by staining normalization; constructing a feature matrix for each pathological image, and further using a two-layer multi-instance learning method to construct a package for each pathological image; constructing a multi-task deep learning network based on transformer and MobileNet; applying the model to a test set and outputting pathological image analysis results. This invention employs a multi-task deep neural network and applies it to the task of predicting multiple gene mutations based on pathological images. Compared to traditional single-task networks, this invention can simultaneously predict the results of multiple tasks, saving computational resources while improving accuracy; furthermore, combining the transformer module with traditional convolution considers both local and global features.
Need to check novelty before this filing date? Find Prior Art