Optimizing PCOD Diagnosis Using Ultrasound Images: Benchmarking VGG and Holonet Models

A prevalent endocrine disorder among post-childbearing women is polycystic ovarian syndrome (PCOD). Symptoms associated with hormonal imbalances include infertility, irregular menstrual cycles, and hirsutism. Timely identification of the disease is essential for effective management and to mitigate associated health concerns. We present a simple approach for diagnosing PCOD utilizing clinical and laboratory characteristics. This approach encompasses pre-processing and segmentation, followed by classification utilizing Holonet. The purpose of this study is to address significant difficulties in the development of the HoloNet deep learning model for accurate PCOD recognition and classification in ultrasound images. These are complex contributions that address the diverse challenges associated with conventional methods to enhance diagnostic precision. HoloNet integrates convolution layers, density blocks, and suitable pooling algorithms, rendering it an exceptionally effective model for accurately capturing and extracting essential information from the input. In all four evaluated criteria—accuracy, recall, F1 score, and precision—Holonet consistently outperformed the competition when compared to VGG 16 and VGG 19, among others. Our research findings indicate that holonet consistently outperforms other models regarding precision, accuracy, recall, and F1 score. As a result, with an accuracy of approximately 98%.