Abstract
BACKGROUND: Ovarian cyst pedicle torsion (OT) has a high clinical misdiagnosis rate, and its early diagnosis largely relies on single ultrasound indicators. This study aimed to integrate multimodal ultrasound parameters (including structural morphology, hemodynamics, and spatial features) and key clinical indicators to construct a risk analysis and prediction model for OT, thereby providing a standardized tool for the early diagnosis of OT. METHODS: The clinical data of 301 patients with ovarian cysts indicated by ultrasound at Gansu Provincial Maternity and Child Health Hospital from September 2021 to December 2023 were retrospectively analyzed. The patients were randomly divided into a training set (n=210, including 101 cases in the OT group and 109 cases in the non-OT group) and a test set (n=91, including 43 cases in the OT group and 48 cases in the non-OT group) at a ratio of 7:3. Multimodal ultrasound parameters were extracted, including structural morphology parameters (cyst diameter, cyst wall thickness, and unilocular/multilocular structure), hemodynamic parameters (pedicle blood flow signals: whirlpool-like/spiral/other types), and spatial feature parameters [cyst location, and heterogeneous nodules adjacent to the cyst (HNACs)]. A clinical indicator (pregnancy status) was also included. Univariate and multivariate logistic regression analyses were performed to identify the independent risk factors of OT, and a nomogram model was constructed. Receiver operating characteristic (ROC) curve, calibration curve, and clinical decision curve analyses were performed to evaluate the model's discriminative ability, calibration performance, and clinical applicability. RESULTS: The multivariate logistic regression analysis showed that the independent risk factors for OT included a cyst diameter of 5-10 cm [odds ratio (OR) =2.768, 95% confidence interval (CI): 2.205-10.872], cyst wall thickness (OR =1.409, 95% CI: 1.119-1.775), whirlpool-like/spiral blood flow in the pedicle (OR =0.26/0.31, 95% CI: 0.08-0.87/0.12-0.97; note: an OR <1 reflects the degree of torsion rather than a protective effect), a HNAC (OR =6.014, 95% CI: 2.315-13.242), a right-sided cyst (OR =4.256, 95% CI: 1.492-12.138), a multilocular cyst (OR =2.555, 95% CI: 1.088-5.999), and pregnancy status (OR =0.316, 95% CI: 0.105-0.954). The nomogram showed excellent discriminative ability [training set: area under the curve (AUC) =0.919, 95% CI:0.877-0.963; test set: AUC =0.876, 95% CI: 0.804-0.949]. Calibration curves revealed good consistency between the predicted probabilities and actual incidence, and the decision curve analysis confirmed the high net benefit across a wide risk threshold range. CONCLUSIONS: Our multimodal ultrasound-based prediction model, which integrates structural, hemodynamic, and spatial parameters, achieved strong internal validation for OT risk prediction. This tool assisted clinicians to rapidly standardize OT risk assessment in this single-center cohort, reducing misdiagnosis, and providing reliable support for clinical decision-making. External multicenter prospective studies are required to further validate its broader applicability.