Evaluation of parametrial infiltration in patients with IB-IIB cervical cancer by a radiomics model integrating features from tumoral and peritumoral regions on 18 F-FDG PET/MRI images.

Abstract

Parametrial infiltration (PMI) is an essential factor in staging and planning treatment of cervical cancer. The purpose of this study was to develop a radiomics model for accessing PMI in patients with IB-IIB cervical cancer using the features from 18 F-FDG PET/MRI images. In this retrospective study, 66 patients with FIGO stage IB-IIB cervical cancer (22 with PMI and 44 without PMI) underwent 18 F-FDG PET/MRI were divided into a training (n = 46) and a testing dataset (n = 20). Features were extracted from both tumoral and peritumoral regions on 18 F-FDG PET/MRI. Single-modality and multi-modality radiomics models were developed with random forest to predict PMI. The performance of models was evaluated with F1 score, accuracy and AUC. Kappa test was used to observe the differences between pathological results and PMI evaluated by radiomics-based models. Intraclass correlation coefficient for features extracted from each ROI was measured. Three-fold cross-validation was conducted to confirm the diagnostic ability of the features. The radiomics models developed by features from primary tumor on T2 weighted (F1 Score = 0.400, accuracy = 0.700, AUC = 0.708, Kappa = 0.211, P = 0.329) and peritumoral region on PET (F1 Score = 0.533, accuracy = 0.650, AUC = 0.714, Kappa = 0.271, P = 0.202) achieved better performances in the testing dataset among four single-ROI radiomics models. The combined model using the features from primary tumor on T2 weighted and peritumoral region on PET achieved the best performance (F1 Score = 0.727, accuracy = 0.850, AUC = 0.774, Kappa = 0.625, P < 0.05). The results suggest that 18 F-FDG PET/MRI could provide complementary information of cervical cancer. The radiomics-based method integrating the features from tumoral and peritumoral regions on 18 F-FDG PET/MRI had a superior performance on PMI evaluation.