Shear wave elastography values in endometrioma: Clinical findings and machine learning-based prediction models.

OBJECTIVE

To evaluate the diagnostic value of shear wave elastography (SWE) in assessing endometriomas and its correlation with clinical symptoms. Furthermore, the study investigates the use of machine learning (ML) models to predict clinical outcomes based on SWE-derived mean shear wave velocity (SWV).

METHODS

This prospective study included 94 women aged 20-35 years diagnosed with unilateral ovarian endometriomas. SWE was used to assess tissue stiffness in meters per second (m/s). Clinical characteristics such as dysmenorrhea, dyspareunia, infertility, and non-cyclic chronic pelvic pain were evaluated. Statistical analyses were performed using SPSS, and ML models, including Logistic Regression, Random Forest, Gradient Boosting, and Support Vector Machine (SVM), were employed to predict clinical outcomes based on mean SWV.

RESULTS

Shear wave elastography-derived mean SWV was significantly higher in patients with dysmenorrhea (4.79 ± 1.08 vs. 3.22 ± 0.65 m/s; P = 0.024), dyspareunia (5.78 ± 1.9 vs. 3.01 ± 0.94 m/s; P = 0.016), and infertility (5.42 ± 1.44 vs. 3.93 ± 1.03 m/s; P = 0.014). ML models demonstrated high predictive accuracy for dysmenorrhea (receiver operating characteristic [ROC]-area under the curve [AUC] 0.94) and dyspareunia (ROC-AUC 0.98), with Logistic Regression and SVM outperforming other algorithms.

CONCLUSION

Shear wave elastography combined with ML offers a non-invasive, cost-effective approach to diagnosing and predicting clinical outcomes in endometriomas. The integration of imaging and computational tools paves the way for precision medicine in gynecology.