Machine Learning based Radiomics from Multiparametric Magnetic Resonance Imaging for Predicting Lymph Node Metastasis in Cervical Cancer

Abstract Background Construct and compare multiple machine-learning models to predict lymph node (LN) metastasis in cervical cancer, utilizing radiomic features extracted from preoperative multi-parametric magnetic resonance imaging (MRI). Methods This study retrospectively enrolled 407 patients with cervical cancer who were randomly divided into training cohort (n = 284) and validation cohort (n = 123). A total of 4065 radiomic features were extracted from the tumor regions of interest on contrast-enhanced T1-weighted imaging, T2-weighted imaging, and diffusion-weighted imaging for each patient. The Mann-Whitney U test, Spearman correlation analysis, and selection operator Cox regression analysis were employed for radiomic feature selection. The relationship between MRI radiomic features and LN status was analyzed by five machine-learning algorithms. Model performance was evaluated by measuring the area under the receiver-operating characteristic curve (AUC) and accuracy (ACC). Moreover, Kaplan–Meier analysis was used to validate the prognostic value of selected clinical and radiomics characteristics. Results LN metastasis was pathologically detected in 24.3% (99/407) of patients. Following three-step feature selection, 18 radiomic features were employed for model construction. The XGBoost model exhibited superior performance compared to other models, achieving an AUC, accuracy, sensitivity, specificity, and F1-score of 0.9268, 0.8969, 0.7419, 0.9891, and 0.8364, respectively, on the validation set. Additionally, Kaplan − Meier curves indicated a significant correlation between radiomic scores and progression-free survival in cervical cancer patients (p < 0.05). Conclusion Machine learning-based multi-parametric MRI radiomic analysis demonstrates a promising performance in the preoperative prediction of LN metastasis and clinical prognosis in cervical cancer.