Classification of Ultrasound PCOS Image using Deep Learning based Hybrid Models

Abstract

Polycystic Ovary Syndrome or (PCOS) is one of the medical disorders that affect female pregnancy. High levels of androgens in women are the root cause of the symptoms that make up Polycystic Ovarian Syndrome (PCOS). According to recent studies, this illness affects roughly 20 of Indian women. Damaged ovaries were identified by a physician’s manual review of ultrasound images, but they were unable to determine whether they were simple cysts, PCOS, or malignant cysts. The majority of imaging characteristics are utilized to diagnose the illness. Ultrasound imaging has become an essential diagnostic technique for PCOS. Because it is essentially an experience-based operation, the typical look of the picture becomes progressively challenging due to overlapping follicles, intrinsic noise of the equipment, and a lack of operator comprehension, making the diagnosis method time-consuming. This paper proposes a method of prediction of PCOS using transfer learning techniques like Alexnet, Inception V3, Resnet50, VGG16 and Hybrid Models. Here, an attempt to offer a methodology in which Hybrid Models will be included in order to train models and improve accuracy. Finally, the test data set is used for the feature extraction procedure and using the performance coefficients, the results are 93 percentage accurate.