The paper presents the first results of a Computer Aided Diagnosis (CAD) system development project for automatic detection of breast cancer. The primary aim of the R&D project is to develop an intelligent advisory system for analyzing mammographic images, and to help radiologists in making diagnosis. One of these results is the set-up of an image library of 10,000 digitalized and qualified mammograms. This work has been started and that will support research and education both in the field of medical sciences and in medical engineering. It is also important that in the frame of the project a prototype of a mammography workstation will be developed and tested. Research tasks are already in an advanced state, algorithms has been developed and tested for microcalcification detection using hierarchical neural nets, and also for lesion characterization using hybrid methods and texture analyses. Current results on a moderate size database are encouraging and we can expect further advances from the utilization of the large image database.