This bacherlor thesis proposes a new paradigm to discover biomarkers capable of characterizing obsessive-compulsive disorder (OCD) by means of machine learning methods. These biomarkers, named neuromarkers, will be obtained through the analysis of sets of magnetic resonance images of the brains of OCD patients and healthy control subjects. The design of the neuromarkers stems from a method for the automatic discovery of clusters of voxels, distributed in separate brain regions, relevant to OCD. This method was recently published by Dr. Emilio Parrado Hernández, Dr. Vanessa Gómez Verdejo and Dr. Manel Martínez Ramón. With these clusters as a starting point, we will de ne the neuromarkers as a set of measurements describing features of these individual regions. Then we will perform a selection of these neuromarkers, using state of the art feature selection techniques, to arrive at a reduced, relevant and intuitive set. The results will be sent to Dr. Carles Soriano Mas at the Bellvitge University Hospital in Barcelona, Spain. His feedback will be used to determine the e cacy of our neuromarkers and their usefulness for psychiatric analysis. The main goal of the project is to come up with a set of neuromarkers for OCD characterisation that are easy to interpret and handle by the psychiatric community. A paper presenting the methods and results described in this bachelor thesis, of which the student is the main author, has been submitted and accepted for presentation in the 2014 European Congress of Machine Learning (ECML/PKDD 2014). The ECML reported a 23.8% paper acceptance rate for 2014.