A Neural Network Approach to MR and CT Image Understanding

The problems encountered in the development of automatic systems for MR and CT image analysis are firstly recalled. An approach based on the integration of artificial neural networks and computer vision techniques which seems to be capable of overcoming such problems is then presented. According to this approach we have developed a system for the construction of 3-D descriptions of the organs imaged in sequences of MR or CT slices. The architecture and the preliminary results of this system are also described.