The prostate of a human being is a very minor organ that plays an essential role in the reproductive system. The production of the fluid in semen that is responsible for transporting sperm throughout the male reproductive system is the major function of the prostate gland. In order for computers to interpret visual stuff, they, like humans, need a collection of facts or qualities that differentiate them from other things. It is preferable to use a programmed scheme to select the greatest diversity of topographies from the data rather than a wide variety of different methods to locate the appropriate ones than to allow the programmed scheme to make those selections. We have decided to adopt a method of deep learning to produce the features that are utilised in prostate segmentation since we want the detection accuracy of our CAD system to be improved. For the initial stage of the segmentation process, the strategy that we have developed uses stacked sparse auto-encoders as the primary tool. The completion of the final segmentation of the prostate is achieved by utilising a deformable model of the system. The precision of the segmentation process has been significantly enhanced as a direct result of the utilisation of deformable models in the process.