Method and system for the automatic recognition of lesions in a set of breast magnetic resonance images

A method of identification of potential lesions of a breast from tomographic image datasets of a chest region of a patient, the- datasets comprising a plurality of voxels (2) each having an intensity value, the images including a region of interest (10) which comprises at least one breast (6). The method comprises the steps of: acquiring a set of images after the administration of a contrast agent to the patient; normalizing (254) the intensity of voxels (2) belonging to the region of interest (10) of the acquired images according to at least one normalization factor; classifying (255) each of the normalized voxels (2) on the basis of a classification criterion, in such a way as to identify regions (40) representing potential lesions. The method is characterized in that the normalization factor is based on normalization voxels (2) corresponding to an anatomical structure (34), the normalization voxels (2) having intensity values enhanced due to the administration of the contrast agent.