This paper presents some possibilities of applying mathematical linguistic formalisms, in the form of structural pattern recognition, in the development of a new class of intelligent medical diagnosis-support and image recognition systems – Cognitive Vision Systems. Such methods are aimed at facilitating the semantic analysis of the meaning of some selected medical patterns based on the presence of cognitive resonance. When trying to apply this cognitive resonance to the automatic analysis and interpretation of image-type data, we propose a new class of information systems: UBIAS (Understanding Based Image Analysis Systems). The procedures proposed are based on the cognitive resonance and categorization model. The application presented in this paper will show how great the opportunities are for automatically detecting lesions in the analysed structures.