Enhanced Detection of Ovarian Cysts Through Texture Features and Rule-Based Techniques

Ovarian cysts are a prevalent condition among women, posing challenges in accurately distinguishing between benign and potentially malignant growths. This study presents a comprehensive diagnostic framework that leverages ultrasound (USG) imaging coupled with advanced classification techniques. The proposed methodology integrates denoising, segmentation, and rule-based classification, with a focus on texture feature analysis to enhance detection and classification accuracy. Initially, USG images undergo pre-processing using advanced denoising filters to improve image clarity. Segmentation of cystic regions is among the techniques employed for this task is the Active Contour Mode, Level Set Method, Graph Cut, Watershed, and Region Growing, ensuring precise boundary identification. Texture features such as entropy, contrast, correlation, and energy are extracted from the segmented regions to aid in distinguishing between cyst types. These features are included into a rule-based classification system to categorize ovarian cysts into Functional Cysts, Dermoid Cysts, Endometriomas, and Serous Cystadenomas, with tailored medical recommendations provided for each type. Quantitative evaluation of segmentation results is carried out using indicators such as MSE, PSNR, and SSIM (Mean Squared Error, Peak Signal-to-Noise Ratio and Structural Similarity Index), ensuring the preservation of cyst structure. By combining texture analysis with robust segmentation techniques, the proposed approach achieves a diagnostic accuracy of 96.45%. This methodology provides clinicians with a reliable and automated tool for ovarian cyst classification, improving diagnostic efficiency and supporting informed medical decision-making.