Integration of radiomics, deep learning, transcriptomics, and metabolomics reveals prognostic risk stratification and underlying biological mechanisms in colorectal cancer - PubMed
5 days ago
- #radiomics
- #colorectal cancer
- #deep learning
- Integration of radiomics, deep learning, transcriptomics, and metabolomics improves prognostic risk stratification in colorectal cancer (CRC).
- A deep learning radiomics model (DLRM) was developed using venous-phase CT images from 1183 patients across four centers, optimized with ten machine learning algorithms.
- High-risk tumors were associated with extracellular matrix (ECM)-related pathways, while low-risk tumors showed immune-related signatures, including higher CD8+ T-cell infiltration.
- Butanoate metabolism and nitrogen metabolism were identified as protective pathways in both omics analyses, validated in an independent cohort.
- The study provides a robust framework for risk stratification and uncovers potential therapeutic biological processes in CRC.