Automated Ovarian Cancer Identification from Biomedical Images using a Multi-Path Deep Learning Framework

Among women's gynecological disorders, cancer of the ovary is one of the most deadly. Tumors that start in the ovaries, whether one or both, are called ovarian cancer. More and more women are being diagnosed with ovarian cancer currently. The death rate from ovarian cancer can be reduced with early identification. Consequently, a lot of recent efforts have focused on developing deep learning models to help detect ovarian cancers from ultrasound pictures, which is essential for diagnosis. Through the utilization of data and algorithmic analytic, machine learning offers a potential solution for ovarian cancer prediction and early detection, which might perhaps defeat this formidable foe. We retrain the characteristics we extracted from CNN using the method known as Random Forest, and then we use the experimental findings from our CNN and Random Forest experiments to make predictions based on OC patient data. This work aims to develop a model that may be used to categorize cases into two groups: those with benign ovarian tumors (BOT) and those with colon cancer (OC). It was found that CNN with Random Forest Model achieved the best accuracy (94.07%), surpassing even the XG Boost model (89%) in our investigation.