| dc.description.abstract | Polycystic Ovary Syndrome (PCOS) a rather common undiagnosed ovarian disorder. It is multifactorial and is diagnosed based on two of three main criteria: ovulatory dysfunction, multiple small cyst presence around ovaries and the high amount of androgen hormones.. PCOS not only causes fertility disorders and irregular menstrual cycles, but also triggers various health problems such as insulin resistance, obesity, hirsutism, acne, hair loss, and the risk of recurrent miscarriages. Given its serious impact on women's reproductive health, early detection of PCOS is crucial. So far, ultrasonography (USG) has been the main modality of PCOS diagnosis due to its ability to directly visualize the number, follicle size, and morphology of the ovaries. However, the interpretation of ultrasound results is still subjective and highly dependent on the expertise of medical personnel. Automation of the detection process can improve the efficiency of diagnosis, allowing medical personnel to treat more patients with more standardized consistency of results. This research aims to develop a You Only Look Once-based system, specifically YOLOv8. The research stages include preprocessing, image augmentation, feature extraction and YOLOv8-based detection in one flow (end-to-end). A total of 1,000 images were employed in the dataset, divided into 800 for training, 100 for validation, and 100 for testing. The evaluation outcomes indicate that the system can recognize PCOS with an accuracy rate of 93%. | en_US |
