Multi-level ensemble for enhanced prognosis of ovarian cancer from histopathological images

Ovarian cancer is a very lethal malignancy that affects women. Early identification of this malignancy is challenging since the first symptoms typically resemble common conditions. The significance of early detection is advantageous. Early diagnosis allows medical professionals to start treatment immediately, thereby improving medical outcomes. We analysed the present situation and built a deep learning-based system for early ovarian cancer identification in women. The design of architecture known as the pyramid ensemble achieved an outperforming 98.18% test accuracy. The training accuracy was 99.57%, with a loss value of 0.0087. Similarly, validation accuracy reached 99.01% with a loss of 0.0072. The computed accuracy, specificity, and sensitivity values were 97.4%, 97.7%, and 98.7%, respectively. Hence, this approach outperforms existing known approaches.