Ultrasound Radiomics and Peripheral Blood Indices Model for Predicting Pregnancy and Recurrence After TCRA in Intrauterine Adhesion.

Objective: The postoperative prognosis (ie pregnancy and recurrence) of patients with intrauterine adhesions(IUA) has always been a concern for women of childbearing age, and there is a lack of prevention and treatment strategies. This study aimed to develop an IUA pregnancy and recurrence (IUA-PR) prediction model to guide clinical decision-making. Materials and Methods: A retrospective analysis was conducted on 387 patients diagnosed with IUA between January 2021 and December 2023. Radiomic features were extracted from ultrasound images using manually designed feature sets, and peripheral blood parameters were integrated with these radiomic features to construct a classification model. The least absolute shrinkage and selection operator (LASSO) combined with the Bayesian information criterion (BIC) was employed to identify nonzero-coefficient features from the radiomic dataset. The predictive efficacy of the developed model was systematically evaluated via the area under the receiver operating characteristic curve (AUC) and decision curve analysis (DCA). Results: A total of five peripheral blood inflammatory indices and six ultrasound radiomic parameters were finally used to construct the IUA-PR prediction model. Among them, the nomogram constructed based on platelet/lymphocyte ratio(PLR), neutrophil/lymphocyte ratio(NLR), aggregate index of systemic inflammation(AISI), ultrasound radiomic score (Rad-score), and postoperative menstrual status showed an AUC of 0.886 for predicting pregnancy outcomes. Additionally, the recurrence prediction model established with systemic inflammatory response index(SIRI), systemic immune-inflammation index(SII), Rad-score, and postoperative menstrual status achieved an AUC of 0.720 in the testing set. Conclusion: We have successfully developed the IUA-PR prediction model constructed based on peripheral blood inflammatory parameters and ultrasound radiomics. Renowned for its convenience and low cost, this model, particularly the generalized linear regression model, exhibits superior predictive performance in forecasting postoperative pregnancy and recurrence, thereby assisting patients in guiding their postoperative fertility decisions.