According to the American Cancer Society, prostate cancer is the second most common cancer and the second leading cause of cancer death among men in the United States. Without an early diagnosis, the chances of related complications such as Lymphoedema, Metastatic Spinal Cord Compression (MSCC), and Hypercalcaemia increase by almost threefold. However, current diagnosis tools are time-consuming, extremely invasive, and result in low accuracy with about an 89 percent false-positive rate. The objective of this study is to provide a non-invasive early diagnosis of prostate cancer by rapidly converting low-dose radiation computed tomography (CT) and magnetic resonance imaging (MRI) scans into superior quality scans. Thus, reducing radiation exposure and increasing the efficiency of diagnosis. The project consisted of developing three main sectors: denoising, generation, and classification. The denoising sector consisted of an AutoEncoder and a CNN (Convolutional Neural Network). Due to limited training data available, a GAN (Generative Adversarial Network) was used to reliably generate more training data, prioritizing the overall efficiency. The GAN was trained on a small portion of the main dataset and drastically optimized the performance of the overall model. The final classification sector consisted of a DCNN (Deep Convolutional Neural Network) for the diagnosis. Overall, these steps resulted in an algorithm that can diagnose prostate cancer with an accuracy rate above 85% all in an accessible and scalable platform, which is a profound improvement over current methods.