DIAGNOSTIC PERFORMANCE OF RADIOMICS-BASED ULTRASOUND ANALYSIS IN WOMEN WITH OVULATORY DYSFUNCTION–RELATED ABNORMAL UTERINE BLEEDING

Abnormal uterine bleeding associated with ovulatory dysfunction (AUB-O) is a common gynecological condition among women of reproductive age. Radiomics enables quantitative analysis of medical images and allows detection of subtle microstructural tissue characteristics. Objective: To evaluate the diagnostic performance of radiomics-based ultrasound analysis for risk stratification of ovarian changes in patients with ovulatory dysfunction–related abnormal uterine bleeding. Methods: A prospective study included 23 women aged 18–45 years with clinically confirmed ovulatory dysfunction. All participants underwent transvaginal ultrasound examination. Images were stored in DICOM format and processed for radiomics analysis. Results: The mean age of patients was 29.6±4.8 years, and the mean body mass index was 24.3±2.9 kg/m². Ultrasound examination identified follicular cysts in 11 cases, lutein cysts in 3 cases, persistent follicles in 5 cases, and suspicious ovarian morphology in 4 cases. Conventional ultrasound classified 16 cases as benign and 7 as suspicious, while clinical follow-up confirmed 14 benign and 9 high-risk cases. A total of 132 radiomic features were extracted; after feature selection, 8 key parameters were included in the model. Significant predictors included entropy (p=0.004), GLCM contrast (p=0.008), homogeneity (p=0.012), and shape irregularity index (p=0.015). Radiomics-based analysis demonstrated higher diagnostic accuracy compared with conventional ultrasound (p