Detection of polycystic ovarian syndrome using follicle recognition technique

Abstract

Polycystic ovary syndrome is a disorder involving prolonged menstrual cycle, and often excess androgen level normally occurs in several women at the time of their reproductive age. This causes impotence along with gynaecomastia and hirsutism. Studying these kinds of condition in women is a major problem which can be resolved by analysing ultrasound images which have the necessary details like number of follicles, size, and position. However, there is a lack of solid objective test that can provide absolute affirmative to diagnose and understand PCOS. This motivates us to think about finding a method to diagnose PCOS at early stages preventing further complications. An automatic PCOS diagnosing tool would help to save the actual time spent on manual tracing of follicles and measuring the geometric features of every follicle. The proposed method was able to achieve classification with accuracy greater than 97% using a KNN classifier. The classifier will improve the time spent on diagonising PCOS and improve its accuracy, reducing the risk of the fatal complications that can be caused by delayed diagnosis.