High resolution MRI-based radiomic nomogram in predicting perineural invasion in rectal cancer

Abstract

Background: To establish and validate a high-resolution magnetic resonance imaging (HRMRI)-based radiomic nomogram for prediction of preoperative perineural invasion (PNI) of rectal cancer (RC). Methods: Our retrospective study included 140 subjects with RC (99 in the training cohort and 41 in the validation cohort) who underwent a preoperative HRMRI scan between December 2016 and December 2019. All subjects underwent radical surgery, and then PNI status was evaluated by a qualified pathologist. A total of 396 radiomic features were extracted from oblique axial T2 weighted images, and optimal features were selected to construct a radiomic signature. A combined nomogram was established by incorporating the radiomic signature, HRMRI findings, and clinical risk factors selected by using multivariable logistic regression. Results: The predictive nomogram of PNI included a radiomic signature, and MRI-reported tumor stage (mT-stage). Clinical risk factors failed to increase the predictive value. Favorable discrimination was achieved between PNI-positive and PNI-negative groups using the radiomic nomogram. The area under the curve (AUC) was 0.81 (95% confidence interval [CI], 0.71–0.91) in the training cohort and 0.75 (95% CI, 0.58–0.92) in the validation cohort. Moreover, our result highlighted that the radiomic nomogram was clinically beneficial, as evidenced by a decision curve analysis. Conclusions: HRMRI-based radiomic nomogram could be helpful in the prediction of preoperative PNI in RC patients.