Prediction of perineural invasion in rectal cancer based on clinical-magnetic resonance radiomics model

YANG Lihong; WANG Binjie; SHI Xiaoying; LI Bairu; YANG Xiaohui; WANG Changfu

doi:10.20258/j.cnki.1006-9011.2025.01.020

Abstract：Objective To establish a clinical-magnetic resonance imaging radiomics model to predict the perineural invasion(PNI)status of rectal cancer before surgery.Methods The preoperative data of 75 patients with rectal cancer were selected,including 26 cases with positive PNI and 49 cases with negative PNI.The Smote algorithm was used to expand the number of PNI-positive patients to 49 cases,and the data from 98 cases were finally included in the analysis.The Pyradiomics software package was used to extract radiomic features from the ADC sequence,and the least absolute shrinkage and selection operator(LASSO)was used to reduce feature dimension calculate the radiomic score(R-score),and an R-score model.Single-factor and multi-factor logistic regression were used to screen the independent risk factors for rectal cancer PNI and was built a clinical model es-tablished.We combined the R-score model with the clinical model to establish a combined model,and analyzed its clinical value through calibration curves and decision curves.Results The AUC of R-score model training set was 0.911(95%CI:0.816-0.966)and the AUC of validation set was 0.837(95%CI:0.657-0.946).Magnetic resonance T stage(mrT)was an independent risk factor for rectal cancer PNI.The AUC of clinical model training set was 0.790(95%CI:0.674-0.879)and the AUC of valida-tion set was 0.701(95%CI:0.507-0.854).The AUC of the combined model training set was 0.952(95%CI:0.871-0.989),and the AUC of validation set was 0.860(95%CI:0.684-0.959).Both the R-score model and the combined model showed good pre-dictive performance,and the fitting curve and clinical decision curve showed that the combined model had good stable perfor-mance and clinical value.Conclusion The clinical-magnetic resonance imaging omics model can non-invasively predict the PNI status of rectal cancer before surgery,providing a basis for individualized clinical decision-making.