Enhancing PCOD Detection in Ultrasound Imaging: an Evaluation of Deep Learning Models

An endocrine disorder that is apparent in many post- childbearing women is polycystic ovarian syndrome, commonly abbreviated as PCOD. Some of the symptoms of the conditions that are associated with hormonal imbalances include infertility couple, anovulation, and hirsutism. This therefore means that there must be early diagnosis of this disease to achieve good control and to avoid the health risks associated with it. Herein below we describe a simple decision making tree for diagnosing PCOD concerning clinical and laboratory abnormalities. This method consists of pre-treatment segmentation and final classification with the support of a holonet Hence, these significant concerns are among the study objectives of this paper in the formulation of a reformed deeper learning model called HoloNet for accurate PCOD recognition and classification in sonography images. The three components, that is Convolution layers, density blocks, and proper pooling algorithms are all integrated into HoloNet making the model uniquely capable of capturing and extracting that essential piece of information from the input. In all four criteria, namely accuracy, recall, F1 score, and precision, the results demonstrate that Holonet outperformed the competition against Resnet 152 and Incepton resnet among others as part of the benchmarking. The benchmark analysis of the performance also reveals that in each of the precisions such as precision, accuracy, recall, and F1 the models of the presented structure of holonet iteratively perform better than the other models. As a result, the use of this technology measures about 98% accuracy.