MRI image plays and important role in medical diagnosis tasks. This work presents a neuroevolutive model for the classification (abnormality/normality) of brain medical digital images to the support and aid to the medical diagnostic performed by specialists. Literature review shows the effectiveness of neural networks in this classification task, our proposal is based on the implementation of a well known genetic algorithm in literature to perform a neuroevolution process, introduced as a optimization to find the best weights on a neural network. The feature extraction process involves a Gabor Filter which perform well and it is rotation invariant, a set of 48 features per images were produced, the neuroevolution is performed at the classification stage using a feedforward multilayer neural network. A set of 210 images were collected and a set of test were performed on dataset and 98.75% of precision was achieved.