Detection of Polycystic Ovarian Disease Using Probabilistic Optimum Deep Learning Feature Fusion
- p1572-1586
Polycystic ovarian disease (PCOD), also referred to as polycystic ovary syndrome (PCOS), is a prevalent disorder affecting most of women between the ages of 12 and 45. A healthy diet and regular exercise are often the first things to fall by the wayside when life gets hectic. Infertility due to polycystic ovary syndrome (PCOS) is frequent and curable. Ultrasound scans with relevant data like the number of follicles, size, and position make a significant contribution to the study of these conditions in women. There is currently no conclusive objective test for the diagnosis or understanding of PCOS. This encourages us to search for a means of early diagnosis of PCOS. One of the difficult PCOS diagnostic criteria is the identification of ultrasound images of the ovaries, which is currently done manually by doctors and radiologists through the counting of follicles and the determination of their volume. We developed a deep-learning solution modified Darknet-53 for diagnosing PCOS from an ultrasound image, and it attained a success rate of 98.81%. Reformed differential evaluation (RDE) and reformed grey wolf (RGW) are two enhanced optimisation algorithms used to choose the best features; A unique probability-based serial strategy is employed to combine the best-selected attributes and categorised using deep learning techniques. The models were put through their paces using a variety of validation techniques, including ROC curve plots, AUC scores, and K-Fold Cross-validation. Finally, we validate our unique technique by applying it to a PCOS dataset procured from the Kaggle repository.
Clinical data Darknet-53 Deep learning Clinical study PCODU ltrasound images