Objective To investigate the clinical application value of radiomics features extracted from preoperative magnetic resonance imaging T2 weighted imaging (T2WI), contrast-enhanced T1 weighted imaging (CE-T1WI) and diffusion weighted imaging (DWI) in predicting lymphovascular space invasion (LVSI) status in patients with cervical cancer. Methods Magnetic resonance imaging data of 254 cervical cancer patients were retrospectively analyzed and randomly divided into a training cohort (n=178) and a validation cohort (n=76) at a ratio of 7∶3. Regions of interest of the lesions were segmented slice-by-slice on axial T2WI, CE-T1WI and DWI single sequences respectively. Radiomics features were extracted using Python 3.8.1 software. Feature selection and dimensionality reduction were conducted using the least absolute shrinkage and selection operator (LASSO) algorithm and recursive feature elimination (RFE) algorithm. A prediction model was established based on the eXtreme Gradient Boosting (XGBoost) algorithm. The area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), F1-score and Brier score were used to comprehensively evaluate the predictive performance and calibration of the model for LVSI in cervical cancer. Results In the training cohert, the AUCs of CE-T1WI, T2WI, DWI single-sequence models and combined radiomics model (Com-Rad) for predicting LVSI status in cervical carcionma were 0.805, 0.842, 0.888, and 0.942, respectively. In the validation cohort, these models for predicting LVSI status in cervical carcionma yielded corresponding AUCs of 0.685, 0.710, 0.730, and 0.853. The Brier score confirmed that Com-Rad demonstrated significantly better calibration ability across both cohorts. Conclusion The Com-Rad radiomics signature developed from preoperative multiparametric MRI features enables accurate prediction of LVSI status in cervical cancer. It provides an objective reference for preoperative clinical assessment of LVSI, optimization of surgical plans and formulation of individualized treatment strategies, and has potential clinical application value.

无