Automated diagnosis of melanomas based on globular and reticular pattern recognition algorithms for epiluminiscence images

In this paper an automated digital system for the diagnosis of melanomas is presented. The system is based on the standard Pattern Recognition dermatological protocol. Two automatic algorithms of digital image processing have been developed in order to detect the appropriate patterns. This allows one to calculate some quantitative features based on the inner patterns of the melanoma using simple-operations algorithms in order to minimize response time. The database used consists of 40 RGB images of 500×500; five (5) of each pattern images have already been catalogued by dermatologists and the results are successful according to the assessment of medical experts. The proposed algorithms presented a remarkable rate of Globular and Reticular Pattern recognition with 0% of false negative and an average of 7.14% of false positive. Thus, it proves to be a reliable system when performing a diagnosis.