Research Neuroinformatics



Artificial intelligence

The application of neural network for professional diagnosis of skin cancer is unique up to today. The capacities of theses learning structures are virtually indefinite.

Because of their ability of adaption neural networks are able to capture and learn the unknown interrelations of the data at hand autonomously. Contrary to the ABCD-rule, where the individual features are in a linear ratio to each other and their weighting is predefined, neural networks are able to to develop and improve such dependencies through an autonomous learning process. Thereby non-linear dependencies and exeptions are taken into consideration.  The DANAOS expert system, for example, concentrates more on the dermoscopic structures of small pigmented skin variances than on the asymetries of the contour.

Early detection of skin cancer per computer / computerised early detection of skin cancer

This unique ability is the reason of the resounding success of neural networks in itelligent data processing. Within the industrial use of this research results up to date the use in the area of recognition of faces and individuals for the safety and monitoring technology has to be stressed above all. Primal the superiority of these self-learning structures versus conventional approaches has made the fascinating development of a worldwide unique tool for the early detection of skin cancer per computer possible.