Utilizing deep learning techniques for detection of Polycystic Ovarian Syndrome using ovarian ultrasound image
Sivakannan Subramani, Aleena Rarichan, S H Chaithra · 2023
This Polycystic Ovarian Syndrome (PCOS) is a medical condition that affects women of reproductive age, characterized by a hormonal imbalance. It usually starts during adolescence, but symptoms may fluctuate over time. PCOS can cause hormonal imbalances due to the cysts present in the ovaries and that can lead to secretion of excess level of androgen. Due to this a PCOS patient can have irregular periods which can cause infertility due to the lack of ovulation. One of the major cause of infertility is PCOS. PCOS is a condition where it is incurable. However, through lifestyle changes, medications and fertility treatments some symptoms can be controlled. Factors like genetics, obesity and high level of androgen can be the causes of the PCOS. But the exact cause is unknown. Polycyst detection using ovarian ultrasonography (USG) scans is a very reliable way to accurately diagnose PCOS and create individualized treatment plans for affected people. Adopting an intelligent computer-assisted cyst detection method is a workable alternative to depending on the imperfect process of manual identification. Thus, we introduced a deep learning-based classification system for PCOS prediction in this research. We train and test the model with a dataset of ovarian ultrasonography (USG) images. To distinguish between ovaries affected by PCOS and those that are not, relevant features from the dataset were specifically extracted using Convolutional Neural Network (CNN). This paper compares different deep learning algorithms such as Alexnet, Resnet50, VGG16 and YOLO V5, which have been found to be the most useful in classifying the ovaries affected by PCOS effectively. The validation techniques show Yolo V5 to be the most effective with an accuracy of 99.8%.