Prognostic risk assessment based on deep learning features of lung cancerpathological images
Principal Investigator
Name
Lei Cao
Degrees
Ph.D.
Institution
Department of Biostatistics, School of Public Health, Harbin Medical University
Position Title
Professor of Biostatistics
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-1397
Initial CDAS Request Approval
Mar 5, 2025
Title
Prognostic risk assessment based on deep learning features of lung cancerpathological images
Summary
The project hopes to automatically extract microenvironmental features of lung cancer pathological images through deep learning methods. Based on these features and weak supervision methods, we can predict downstream tasks of clinical interest, including molecular classification prediction and prognostic risk assessment. The research content of the project mainly includes: representation learning methods based on graph neural networks, feature aggregation methods based on weak supervision or semi-supervision, and construction of prognostic risk markers.
Aims
1. Extracting microenvironmental features based on pathological images
2. Test the performance of the benchmark weakly supervised method on NLST data
3. Developing a prognostic marker construction model based on graph neural networks
4. Use markers to assess patient prognostic risk and analyze their statistical and clinical significance
Collaborators
Liuying Wang, Department of Health Management, Harbin Medical University