Automatic Screening Systems
Making efficient automatic clinical screening systems has been a serious requirement claimed worldwide long time ago. The motivation is that by applying screening systems the number of expensive doctor-patient visits could be reduced and also developing countries could have access to such tools, where there is no sufficient medical care. We wish to focus on developing screening systems in a field where users with the help of image capturing devices can take shots on their own which they can forward through the Internet for evaluation. Such special domains are e.g. the complication of diabetes affecting the eyes, or the presence of skin cancer (melanoma). Higher precision of detection can be assured by using complex (ensemble-based) systems; however, they demand rather high costs. The improvement of mass screening programmes and that of the resolution of image processors foreshadow the significant growth of data. Because of this challenge, we plan to develop distributed image processing algorithms, and also the extension of ensemble-based systems to distributed environment (e.g. for finding optimal parameter settings). One can expect a further significant improvement of the screening systems’ precision due to the inclusion of proteomic data besides image data. In order to achieve the aim we set, machine learning systems handling Big Data and data mining systems are needed.
