Study On The Value Of CT Imaging Model In Predicting Peritoneal Metastasis Of Ovarian Cancer

Objectives:To investigate the value of CT imaging model in predicting peritoneal metastasis of ovarian cancer.Materials and methods:The clinical data and image images of 124 patients with ovarian cancer confirmed by surgical pathology in Jiangxi Provincial People’s Hospital and the First Affiliated Hospital of Nanchang University from June 2014 to May 2022 were retrospectively collected,among which 79 patients with positive peritoneal implant metastasis and 45 patients with negative metastasis were randomly divided into training group(n=88)and verification group(n=36)according to the ratio of 7:3.Special image processing software(ITK-SNAP version 3.6)was used to delineate Three-dimensional(3D)tumor regions of interest on plain CT images.The images were imported into AK(Artificial Intelligence Kit,Version 3.0.1A)software to extract the texture features of lesions.The maximum correlation minimum redundancy(MRMR)criterion and LASSO regression were used to screen the extracted image omics features.The multi-factor logistic regression method was used to train the selected features,and the3 D image omics model was constructed.Area under curve(AUC)was used to evaluate the model’s performance in predicting peritoneal metastasis of ovarian cancer.Results:The eight most predictive radiomics features were selected from the texture features,combined with clinical predictors,and jointly constructed the radiomics nomo model,which showed better predictive ability than the radiomics model and clinical model alone,with an AUC of 0.91,a 95% CI of 0.85-0.97 in the training group,0.82 in the validation group,and 0.67-0.97 in the 95% CI,respectively.Decision curve analysis(DCA)also verifies the superiority of radiomics nomograms.Conclusions:The Normogram model based on the 3D imaging omics and clinical features of plain CT scan has a good ability to predict the peritoneal metastasis of ovarian cancer.