Classification of PCOS Using Machine Learning Algorithms Based on Ultrasound Images of Ovaries

Polycystic Ovary Syndrome is a metabolic endocrine lifestyle disorder. It can be found in women after their puberty. Multiple number of follicles occurred in the ovaries which leads to irregular in menstrual cycle, imbalance BMI/Obesity, numerous hair fall, unwanted hair growth in face and other parts of the body. The above issue influences the women to a risk of infertility and also causes diabetes type II, heart disease, chances for ovarian cancer and endometrial cancer, etc. Main objective of this paper is to classify the whether the women is affected with PCOS or not without any doctors supervision. To diagnose that, the machine has to analyze the ultrasound images of ovary contains the PCOM or not. In order to identify, classification is done based on finding the similarity images an, analyze the minimum distance, finding the edges and contour of an image. In existing systems classification is performed with minimal number of datasets. This paper uses a large amount of datasets for training the images thereby giving a better PCOM results. Classification algorithm such as Support Vector Machine (SVM), K-Nearest K-neighbor (K-NN) and Convolution Neural Network (CNN) algorithm used thereby the conclusion shows that CNN algorithm gives an accuracy of 0.99.