Prostate cancer is the second most common cancer among men. Multiparametric magnetic resonance imaging is a well-established tool for prostate cancer detection and diagnosis. However, image interpretation is complex, time-consuming and subject to high inter-reader variability. In this work, artificial intelligence methods are used to build a computer-aided detection system able to automatically discriminate between healthy and cancer-positive patients. In particular, a deep learning approach based on convolutional neural networks has been developed. The final result shows that the designed model presents good classification performances and is therefore well suited as a decision support tool. This thesis is the result of a seven-month stay at the Norwegian University of Science and Technology (NTNU), Trondheim, under the CIMORe group led by Professor Tone Frost Bathen..