Presenter Information

David Erdos


Electrical and Computer Engineering


Electrical Engineering

Research Advisor

Stanley, R. Joe

Advisor's Department

Electrical and Computer Engineering


Melanoma lesions that exhibit irregular linear vessels can be misdiagnosed due to the fact that their clinical appearance can often closely resemble other skin conditions. Detection these irregular linear blood vessels, or telangiectasia could improve the accuracy of melanoma diagnosis. An algorithm that detects telangiectasia has been developed; it initially searches for color drops in an image in various directions and then uses various post processing techniques to improve the accuracy. The feasibility of using classifiers such as neural networks to further improve the accuracy has also been investigated.

Presentation Type

Oral Presentation

Document Type