Polycystic Ovary Syndrome (PCOS) -Associated Infertility: Clinical Insights and Emerging Deep Learning Approaches for Early Detection and Prediction

Abstract

Polycystic ovary syndrome (PCOS) is the most common endocrine-metabolic disorder in women of reproductive age and a major cause of anovulatory infertility around the world. It is a heterogeneous disorder characterised by chronic oligo- or anovulation, clinical and/or biochemical hyperandrogenism, and polycystic ovarian morphology, often complicated by insulin resistance and obesity. This review highlights current understanding of the epidemiology, pathophysiology, diagnostic criteria, mechanisms of infertility due to PCOS, and management of PCOS patients. Due to significant inter-observer variability in ultrasonography assessment of the polycystic ovarian morphology, more and more research work is being done in using deep learning (DL) algorithms such as convolutional neural networks (CNN) and DL-based architecture including InceptionV3, ResNet, VGG16, and DenseNet to automatically diagnose the polycystic ovarian morphology and predict PCOS risk from clinical and hormonal data . In this review we discuss DL architectures used in studies, the datasets, their performances, and limitations, and future directions of using artificial intelligence in infertility in PCOS.