Ultrasound-Based Ovarian Cysts Detection with Improved Machine-Learning Techniques and Stage Classification Using Enhanced Classifiers

• Published in: SN computer science Vol. 4; no. 5; p. 571
• Main Authors: Athithan, Senthil, Sachi, Savya, Singh, Ajay Kumar
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.09.2023; Springer Nature B.V
• Subjects: Abdomen; Algorithms; Amenorrhea; Artificial neural networks; Blood tests; Breast cancer; Classification; Classifiers; Computer Imaging; Computer Science; Computer Systems Organization and Communication Networks; Congenital diseases; Cysts; Data Structures and Information Theory; Decision support systems; Eggs; Female reproductive system; Females; Fertility; Follicles; Infertility; Information Systems and Communication Service; Machine Intelligence and Smart Systems; Machine learning; Mammography; Original Research; Ovarian cancer; Ovulation; Patients; Pattern Recognition and Graphics; Polycystic ovary syndrome; Radiation; Reproductive health; Software Engineering/Programming and Operating Systems; Support vector machines; Tumors; Ultrasonic imaging; Uterus; Vision; Womens health; Demographic and diagnostic features; Medical diagnostics; Image processing; Ovarian detection; Machine learning; Texture features; K-means clustering
• ISSN: 2661-8907; 2662-995X; 2661-8907
• DOI: 10.1007/s42979-023-01973-0

Infertility is a global issue. Total fertility rate declined from over 5 live births per woman in 1950–1955 to 2.5 births per woman in 2010–2015. Female infertility is 37% globally and 12.5% in India. Female infertility can be reduced by identifying and treating the underlying cause early. Ovulatory diseases contribute 25% globally to female infertility. Various female pelvic imaging methods can diagnose ovulatory problems. Diagnostic ultrasound is chosen because it is radiation- and contrast-free and cost-effective. The intensity-based grouping and textural data were used for the detection of follicles and cysts in the ovary, which is based on machine learning (ML). Ovarian diagnosis was given a major boost thanks to the application of machine-learning algorithms, which permitted a success rate of 97% and significantly improved the overall quality of the process. Standard machine-learning strategies have been looked into for the purpose of ovarian classification. In order to determine which method of classification produces the most accurate findings, we constructed three distinct models utilising artificial neural networks, discriminant classifiers, and support vector machines. When the results of the various created classifiers were compared, it was discovered that SVM had the highest accuracy (98.5%). This ingenious tool, which may be classified as a decision support system, will assist the attending physician in reaching the appropriate determination and preventing an error in his or her interpretation.