Machine learning techniques for medical images in PCOS

Polycystic Ovary Syndrome is an endocrine and lifestyle disorder that affects women of the reproductive age. It is a disorder that causes the ovaries to enlarge due to the presence of cysts. PCOS can be detected by using various techniques like blood tests, pelvic examination and ultrasound imaging. These clinical methods are time consuming, costly and the results generated by them are also prone to human errors. These issues can further lead to low efficiency in the generated results and also hamper in the diagnosis of PCOS. Therefore an automated system for the detection of PCOS can be used to overcome these limitations. An automated system can be useful for medical practitioners to detect PCOS in less time and hence reducing the risk of human errors. In our approach to the detection of PCOS, artificial intelligence techniques which include machine learning, transfer learning and various deep learning techniques play a very important role. For the diagnosis of PCOS various AI techniques are used like segmentation, classification and detection through various classifiers such as SVM, KNN, Logistic Regression etc. This paper reviews existing works in which segmentation, detection and classification is done simultaneously to create an automated system for diagnosis of PCOS through the dataset of ultrasound imaging.