Objective To draw a radiomics nomogram based on CT venous phase imaging and explore its clinical value in distinguishing benign and malignant epithelial tumors of ovary.

Methods A total of 221 patients with pathologically confirmed benign and malignant ovarian epithelial tumors were retrospectively collected, including 86 benign and 135 malignant cases.They were randomly divided into a training group (154 cases) and a validation group (67 cases) in a 7∶ 3 ratio.The region of interest was manually delineated on the enhanced CT venous phase images of each patient, and the radiomics features were extracted.Feature screening was performed and the radiomics model was constructed, and the radiomics label score Rad-score was calculated.Combining clinical factors and radiomics features, a multiple logistic regression model was used to construct a joint model and create a nomogram.The clinical application value of the joint model was evaluated through ROC curve, calibration curve, and decision curve analysis.

Results Tumor imaging showed unclear boundaries, cystic and solid tumors, and CA125 was an independent risk factor for ovarian epithelial malignancy (P < 0.05).Five venous phase features were selected, and the diagnostic efficacy showed an AUC of 0.893 in the training group and 0.869 in the validation group.The prediction results of the imaging radiomics combined model had higher consistency with the pathological results.

Conclusions Establishing a combined model based on CT venous imaging radiomics features and clinical factors, and constructing a nomogram, can intuitively and accurately distinguish between benign and malignant ovarian epithelial tumors.