Skin Lesion Recognition using Deep Neural Networks

OData support
Supervisor:
Szemenyei Márton
Department of Control Engineering and Information Technology

One of the most exciting developement trends in the 21th century is artificial intelligence and its use in both the industrial and consumer segments. The extraordinary increase in computing capacity allows more complex and sophisticated algorithms to be created, while the more intense use of sensors in smart devices and the ever growing scale of information collection means that data amount is no longer a limiting factor. The use of deep neural networks in medical image processing also has a huge potential, as the algorithms can recognise patterns in patients' data which would be otherwise overlooked by doctors. This can raise the quality of diagnostics and medical care, reducing the error rate and the number of unnecessary tests. In my Masters Thesis, I create a neural network to automatically recognise melanoma, one of the most deadly skin cancer types. I also discuss the traditional methods of recognition, comparing the pros and cons of the artificial intelligence method against the traditional way.

Downloads

Please sign in to download the files of this thesis.