Prediction of intermediate and high risk factors based on MRI imaging radiomics before neoadjuvant therapy for cervical cancer

Objective: A fusion model based on axial LAVA+C sequence images and clinical characteristics on preoperative MRI before neoadjuvant therapy is discussed for identifying intermediate and high risk factors in cervical cancer patients. Methods: In a retrospective analysis of 145 cervical cancer patients treated at Fujian Cancer Hospital between January 2013 and July 2018, cases with more than two intermediate risk factors or more than one high-risk factor were classified as positive based on pathological findings, while cases with fewer than two intermediate risk factors and no high-risk factors were classified as negative based on pathological findings. Using an entirely random process, the cases were split into 116 cases for the training set and 29 cases for the test set, based on the Ax-LAVA+C sequence to extract radiomics features. dimensionality reduction for features LASSO and spearman correlation coefficient are used to select the most advantageous radiomics features. The best radiomics models may be filtered out using seven machine learning techniques. To create a composite radiomics model, combine the imaging radiomics model and the clinical model. And assess the model's effectiveness using the ROC, decision curves, and calibration curves. Results: The AUC of the validation set in the clinical- radiomics model was 0.823, the accuracy was 0.793, the sensitivity was 0.667, and the specificity was 0.972, which were higher than the clinical model and comparable to the radiomics model. Conclusion: Before beginning neoadjuvant therapy, the MRI radiomics model based on the Ax-LAVA+C sequence is effective in identifying intermediate- and high-risk variables for postoperative cervical cancer.