Ovarian Cancer is the fifth leading cause of cancer death in women and is the most important type of cancer to study of the female reproductive system.With the application of Deep Learning Models, Early Detection problems have significantly helped the medical community to provide right diagnosis and treatments.However, not all cancer types are given equal importance, especially since ovarian cancer does not have 100% accurate diagnosis test, it becomes highly important to throw some light regarding the problem.
The main contribution of this current work is to make accurate predictions for diagnosis and mortality based off of the two widely used blood scans called the CA-125 and Transvaginal Ultrasound (TVU).Since, Biopsy or Laparactomy act as an ultimate indication for diagnosis, the main inspiration to study the first prediction was if it was possible to make predictions using the parameters from the blood scans and secondly if we could also mae some predictions regarding the mortality of the patients.
The main idea was to build neural network models, train them using different configuration to make predictions.Accordingly, five models were built, cross validated.
The accuracy was then improved by applying the Consensus Algorithm.
The work also interpreted features to understand its contribution. The data was obtained from the National Cancer Institute and some of the results and code is provided in the repective folders.