Novel magnetic resonance imaging (MRI)-based radiomics for predicting perineural invasion in rectal cancer: a two-centre study.

AIM: This study investigates the use of multiparametric magnetic resonance imaging (mp-MRI)-based radiomics for assessing perineural invasion (PNI) in rectal cancer.

MATERIALS AND

METHODS

A retrospective analysis was performed on clinical and MRI data from 423 rectal cancer patients with confirmed surgical pathology, gathered from two centres. Of these, 343 patients from centre 1 were divided into a training set and an internal validation (in-vad) set in an 8:2 ratio, while 80 patients from centre 2 were used for independent external validation (ex-vad). Univariate and multivariate analyses were conducted on clinical features to build a clinical model. A combined model integrating both clinical and radiomic features was developed.

RESULTS

Among all patients, 131 cases (31.0 %) were PNI-positive. A multivariate analysis revealed MRI-reported T (mrT) stage (odds ratio [OR] = 1.66, P=.010) and MRI-reported N (mrN) stage (OR = 1.91, P=.002) as independent predictors of PNI, forming the clinical model. After selecting radiomic features, 30 features were used to construct the radiomics model. The area under the curve (AUC) values for the clinical model in the training, in-vad, and ex-vad sets were 0.719, 0.631, and 0.760, respectively. The AUC values for the radiomics model in the training, in-vad, and ex-vad sets were 0.841, 0.815, and 0.916, respectively, while the AUC values for the combined model in the training, in-vad, and ex-vad sets showed AUC values of 0.899, 0.826, and 0.914, respectively.

CONCLUSION

The mp-MRI-based radiomics model demonstrates high accuracy in predicting PNI status in rectal cancer, offering a noninvasive and reliable tool for preoperative assessment.