Prostate Cancer Detection Using Integrated Multi Modal Approaches

Cancer is one of the terminal diseases that significantly affects an individual's health. Tumours are created when aberrant cells proliferate and divide out of control, posing a threat to neighbouring tissues and organs. Early detection through regular screenings and check-ups can make cancer more treatable. Lots of people throughout are impacted by prostate cancer, the cancer in males that is most frequently diagnosed. The proposed model is a multimodal hybrid approach that utilizes data from .csv files containing prostate attributes and prostate images. It employs both ML and DL models to classify the data as cancerous or non-cancerous and can further stage the cancerous cells into four stages. The proposed method uses the majority voting system to identify the cancerous or non-cancerous. The model achieves an overall accuracy of 81.48 %, with the ML model at 77.8% and the DL model at 87%.