Abstract
Due to advances in the field of machine learning the use of automatic diagnostic imaging systems for the diagnosis of pigmented skin lesions, especially for melanoma, is increasing. These systems work based on a process usually involving the following four steps that build on each other: 1) Pre-processing, a way to assure that the images are without any artifacts that would interfere with the analysis, 2) segmentation, the task of dividing the image into the lesion and its background, 3) extracting features from the image and selecting the ones most useful for an accurate diagnosis, and finally, 4) Classification, the last step of the diagnosis, in which the lesion is assigned to a class. Recently, applications for automated diagnoses of melanoma are based on transfer learning, which makes step 2 and 3 (segmentation and feature extraction) unnecessary and provides better results.

Bio-Sketch
Actual position: Senior physician, Department of Dermatology, Research team leader: In vivo skin imaging group
Scientific profile: H-index (according to ISI): 36 Publications in peer reviewed journals: 160 (55 as first or last author)
Reviewer for high-impact scientific journals: Lancet Oncology, Archives of Dermatology. Journal of the American Academy of Dermatology. Journal of the European Academy of Dermatology and Venerology. Acta dermatologica venerologica. American Journal of Dermatopathology
 
Reviewer for the National Institute of Health (NIH), grant applications
Member of the editorial board of 2 international scientific journals: BMC Dermatology Dermatopathology: Practical and Conceptual
Editor in chief of an international scientific journal Dermatopathology: Practical and Conceptual 2009-2013

Teaching profile:  Editor and author of 2 books, author of 8 book chapters 50 invited lectures in the last 7 years (international) Web based learning and online education (lecturer and editor) Co-editor of DERM101 Director of national and international academic courses

https://cluster.meduniwien.ac.at/mic/