Comparative evaluation of the diagnostic performance of radiomics-based ultrasound analysis in the differential diagnosis of ovarian cysts

Ovarian cysts are among the most common gynecological conditions in women of reproductive age. Ultrasound examination remains the primary imaging modality for evaluating adnexal masses; however, conventional ultrasound interpretation is often dependent on the subjective assessment of the examiner. In recent years, radiomics-based image analysis has emerged as a promising approach for extracting quantitative imaging biomarkers that may improve diagnostic accuracy. This study aimed to comparatively evaluate the diagnostic performance of radiomics-based ultrasound analysis and conventional ultrasound examination in the differential diagnosis of ovarian cysts. Methods. A prospective observational study included 24 female patients aged 19–45 years with sonographically confirmed ovarian cysts. All patients underwent clinical evaluation, gynecological examination, and transvaginal ultrasound with Doppler assessment. Ultrasound images were processed using MATLAB and Simulink, followed by radiomics feature extraction with the PyRadiomics platform after manual segmentation of the region of interest (ROI). Results. Functional cysts (follicular and corpus luteum) accounted for 66.7% (16/24) of cases, while neoplastic cysts (dermoid and endometriotic) represented 33.3% (8/24). A statistically significant difference in cyst distribution was observed (χ²=4.03; p=0.045). Radiomics analysis revealed characteristic imaging patterns for functional cysts, including anechoic content, smooth margins, and high internal homogeneity, which corresponded with histopathological findings. Neoplastic cysts demonstrated higher entropy values and heterogeneous texture patterns. Conclusion. The integration of radiomics into routine ultrasound imaging may enhance diagnostic precision, reduce subjectivity in image interpretation, and support more accurate clinical decision-making in gynecological practice.