Multicenter evaluation of radiomics and multimodal artificial intelligence for colorectal cancer diagnosis and survival prediction
Reliable preoperative classification and outcome prediction could improve management of locally advanced colorectal cancer. We retrospectively studied 879 participants from two centers, including 741 patients with locally advanced colorectal cancer and 138 with benign colorectal lesions. Clinical, radiomics, deep learning, and early fusion models were trained at center A ( n = 615) and independently tested at center B ( n = 264) using pretreatment contrast-enhanced computed tomography and clinical data. The clinical-radiomics nomogram achieved a test AUC of 0.957 for benign-malignant classification. For 3-year and 5-year overall survival, radiomics achieved test AUCs of 0.874 and 0.898, while fusion achieved 0.870 and 0.899, respectively. These findings identify CT radiomics as a promising digital biomarker for colorectal cancer stratification and show that early deep learning fusion offers limited incremental value over an optimized radiomics baseline.
Authors
- Chenquan Gan (ORCID: https://orcid.org/0000-0002-0453-5630)
- Qi Ding (ORCID: https://orcid.org/0009-0004-2708-3210)
- Ketong Wu
- Yong-Qing Ye (ORCID: https://orcid.org/0000-0002-3852-0701)
- Long Xia
- Cheng Li
- Yawen Cao
- Zixuan Zhou
- Jiaxin Ying
- Wenfeng Xie
- Jiawei Zhou
- ChoyYee Tham
- Rongqi Li
- Guodong Pang
Institutions
- Chongqing University of Posts and Telecommunications (CN)
- Guangzhou University of Chinese Medicine (CN)
- Sun Yat-sen University (CN)
- University of Macau (MO)
- Third Affiliated Hospital of Southern Medical University (CN)
- Sixth Affiliated Hospital of Sun Yat-sen University (CN)
- Foshan Hospital of TCM (CN)
- Inner Mongolia People's Hospital (CN)
- Foshan Second People's Hospital (CN)
- Hohhot First Hospital (CN)
Publication Details
- Journal
- npj Digital Medicine
- Published
- 2026-11-19
- DOI
- https://doi.org/10.1038/s41746-026-03434-3
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 0.00