Related Experiment Video
Updated: Aug 15, 2026
Competing-Risk Nomogram for Predicting Cancer-Specific Survival in Multiple Primary Colorectal Cancer Patients after Surgery
Published on: September 27, 2024
Can Multiomics Modeling Enable Accurate Prediction of Microsatellite Instability in Colorectal Cancer?
Weiqun Ao1, Yijiang Huang2, Guoqun Mao3
1Department of Radiology, Tongde Hospital of Zhejiang Province, Hangzhou, Zhejiang, China (W.A., G.M., S.D.); Zhejiang Academy of Traditional Chinese Medicine, Hangzhou, Zhejiang, China (W.A.).
A novel multiomics deep learning model accurately predicts microsatellite instability (MSI) in colorectal cancer (CRC) using CT radiomics and pathomics. This noninvasive approach aids in molecular stratification and personalized treatment for CRC patients.
Area of Science:
- Oncology
- Radiology
- Computational Pathology
Background:
- Microsatellite instability (MSI) is a crucial biomarker in colorectal cancer (CRC) for predicting treatment response.
- Accurate preoperative prediction of MSI status is essential for personalized therapeutic strategies.
Purpose of the Study:
- To develop and validate a multimodal deep learning model for predicting MSI status in CRC.
- To integrate computed tomography (CT) radiomics and histopathologic pathomics for enhanced predictive accuracy.
Main Methods:
- Retrospective analysis of 509 CRC patients from two medical centers.
- Extraction of deep learning features from CT and H&E-stained images using ResNet-101.
- Development of CT deep learning radiomics score (DLRS), pathomics score (DLPS), and a multiomics nomogram integrating clinical variables.
Main Results:
- The multiomics nomogram achieved high predictive performance with AUCs of 0.996 (training), 0.999 (internal validation), and 0.993 (external validation).
- Individual models (DLRS, DLPS, preoperative) showed AUCs ranging from 0.919 to 0.963, outperforming the clinical model (AUCs ~0.75-0.79).
Conclusions:
- The integrated deep learning nomogram provides accurate, noninvasive prediction of MSI status in CRC.
- This multiomics approach serves as a promising imaging-derived biomarker for molecular stratification and personalized treatment decisions in CRC.
