Contrast-enhanced CT-based radiomics nomogram for the preoperative prediction of perineural invasion in colorectal cancer: a two-center study

Objectives/Purpose To develop and validate a radiomics nomogram for preoperative prediction of perineural invasion (PNI) in colorectal cancer (CRC) using multi-center contrast-enhanced CT datasets. Methods This retrospective study enrolled 334 CRC patients from two hospitals, divided into the training, internal validation, and external validation sets. Radiomics features were extracted and selected by the least absolute shrinkage and selection operator (LASSO) logistic regression to build a radiomics model. Clinical independent risk factors were used to construct a clinical model. An ensemble model was established by integrating the Rad-score with clinical predictors. Model performance was evaluated using ROC curves and decision curve analysis (DCA). Results The radiomics model was built from whole-tumor radiomics features, while the clinical model incorporated CEA, CT-reported T stage, and lymph node status. The ensemble model achieved the best performance in the internal validation set, whereas the radiomics model demonstrated greater stability in the external validation set. A nomogram combining the Rad-score and clinical predictors was developed for preoperative PNI prediction. Conclusions The contrast-enhanced CT-based radiomics nomogram is a promising non-invasive tool for preoperative PNI assessment in CRC patients, showing robust performance across multi-center datasets.