Analysis of Ultrasound Parameters Influencing Endometrial Receptivity and a Pregnancy Outcomes Predictive Model for Patients Undergoing In Vitro Fertilization and Embryo Transfer(IVF-ET): A Prospective Study

Chunlian Wang^1*Jiao Lin¹Xingping Zhao²Ni Liu¹Pengzi Sun¹Jiaoli Yang¹Xue Zhou¹

• ¹Xiangtan Central Hospital, Xiangtan, China
• ²The Third Xiangya Hospital of Central South University, Changsha, China

Abstract Purpose: This study aims to assess the impact of ultrasound parameters on endometrial receptivity in patients undergoing in vitro fertilization and embryo transfer (IVF-ET) and to establish a predictive model for ongoing pregnancy outcomes. Methods: A prospective cohort study was performed, enrolling 86 patients treated at the Reproductive Center of Xiangtan Central Hospital from May to December 2024. Participants underwent comprehensive ultrasonographic evaluations, including color Doppler ultrasound, three-dimensional power Doppler angiography (3D-PDA), and endometrial contrast-enhanced ultrasound (CEUS), on the day prior to embryo transfer. Data on endometrial morphology, blood flow characteristics, and parameters derived from 3D-PDA and CEUS were analyzed using lasso regression to construct a predictive model for ongoing pregnancy. Results: Of the 86 patients, 42 (48.8%) achieved ongoing pregnancy, while 44 (51.2%) did not. Significant intergroup differences were noted in the number of mature oocytes (MII oocytes) and endometrial blood flow grading (both P < 0.05). Lasso regression identified eight predictive variables: primary cause of infertility, baseline luteinizing hormone (LH) levels, number of MII oocytes, uterine cavity volume, endometrial blood flow grading, subendometrial flow index (FI) in 3D-PDA, and endometrial and subendometrial peak intensity (PI) in CEUS. These variables were integrated into eight machine learning models, with the Gradient Boosting model exhibiting superior predictive performance (AUC-ROC: 0.980; 95% CI: 0.951–1.000). SHapley Additive exPlanations (SHAP) analysis indicated that a higher number of MII oocytes, improved endometrial blood flow, specific infertility etiologies, elevated baseline LH levels, and reduced subendometrial/endometrial PI, subendometrial FI, and uterine cavity volume were associated with a greater likelihood of pregnancy. Conclusion: The developed predictive model demonstrates robust efficacy in forecasting ongoing pregnancy in IVF-ET patients. Refinement of treatment strategies, including optimization of MII oocyte yield, endometrial blood flow grading, and endometrial/subendometrial PI and FI, may further enhance pregnancy success rates.