DeepOvaNet: A Comprehensive Deep Learning Framework for Predicting and Diagnosing Ovarian Cancer in Women Across Menopausal Transitions

Ovarian cancer is a type of cancer that begins in the ovaries, the female reproductive organ that produces eggs. It is the fifth most common cause of cancer-related death among women. It is more commonly diagnosed in women who have gone through menopause, typically around the age of 50 years or older, that is, across the menopausal transition. This study aimed to evaluate the effectiveness of convolutional neural network (CNN) models in detecting ovarian cancer by examining histopathological images. The evaluation of the performance of the 18 CNN models in differentiating between malignant and non-cancerous histological pictures involved executing each model independently 20 times. The performance of the models was assessed by employing several metrics derived from the confusion matrix, including the accuracy (Acc.), sensitivity (Sen.), specificity (Spec.), Precision (Prec.)., F1 score, false- positive rate (FPR), Matthews Correlation Coefficient (MCC), Kappa, and Computational time. The darknet19 model had superior performance compared to all other models, with an average accuracy of 99.79%, minimum accuracy of 98.95%, and maximum accuracy of 100%. Additionally, the confusion matrix exhibited the following mean values: sensitivity (Sens.) of 99.73% and specificity (Spec.) 99.84% precision (Prec.) of 99.84%, false positive rate (FPR) of 0.15%, F1 score of 99.79%, Matthews correlation coefficient (MCC) of 99.58%, Kappa coefficient of 99.58%, and computation time of 9.58 seconds. In the future, deep learning may be employed to improve the identification of ovarian cancer subgroups.