Existing embedded devices for skin examination have insufficient quality in detecting melanoma or do not support operation by laymen. In addition, there is a lack of security features for accidental hardware failures and external attacks.
This subproject addresses the aforementioned problems through research on an embedded device for skin examination that combines image-based sensor technology with radar sensor technology. The collected data serves as input to neural networks for melanoma detection, with appropriate AI resources (e.g., FPGA) of the embedded device providing energy efficiency and real-time processing. Built-in security features include fault tolerance against random hardware failures, safety monitoring, and security mechanisms against external attacks.
Communication interfaces allow secure data exchange with a physician. Power management services enable long battery life of the device. Sensor technology and algorithms for detecting operating errors and deriving recommended actions provide support for the use of the device by laypersons.
The results are an important input for the overall project for the realization of melanoma detection and telemedicine applications, and also for collaboration evaluation with medical partners. The embedded device with the associated diagnostic algorithms is an effective basis for certifiable melanoma detection and a foundation for telemedical integration.
The participating research group at the University of Siegen has the experience and expertise for the research on the embedded device. Aspects of fault tolerance, energy efficiency, diagnostics, security and artificial intelligence in embedded devices have been investigated in numerous projects.