Enhancing Prostate Cancer Diagnosis with Multi-class Classification of CT Scan Images

This work intends to develop a multi-class classification system for prostate cancer prediction utilising CT scan data in order to enhance prostate cancer diagnosis and treatment planning, and hence patient outcomes. To begin, a preprocessing phase in this research eliminated noise and artefacts from the CT images. Following that, we extracted features using a hybrid approach by mixing deep learning with classic image processing techniques. Following the retrieval of features, they were used to train a convolutional neural network (CNN) multi-class classification model. The technique's efficiency was evaluated experimentally using a collection of CT images from males diagnosed with prostate cancer. The results demonstrated that the proposed multi-class classification approach was very accurate in predicting the existence of prostate cancer as well as the stage of the disease. In conclusion, our work shows that multi-class classification models may be used to predict prostate cancer using CT scan pictures. The described strategy has the potential to significantly increase both the precision with which prostate cancer is identified and the efficacy with which treatment strategies are devised.