Multifaceted analysis for medical data understanding: from data acquisition to multidimensional signal processing to knowledge discovery

Abstract

Large quantities of medical data are routinely generated each day in the form of text, images and time signals, making evident the need to develop new methodologies not only for the automatization of the processing and management of such data, but also for the deeper un- derstanding of the concepts hidden therein. The main problem that arises is that the acquired data cannot always be in an appropriate state or quality for quantitative analysis, and further processing is often necessary in order to enable automatic processing and manage- ment as well as to increase the accuracy of the results. Also, given the multimodal nature of medical data uniform approaches no longer apply and specific algorithm pipelines should be conceived and devel- oped for each case. In this dissertation we tackle some of the problems that occur in the medical domain regarding different data modalities and an attempt to understand the meaning of these data is made. These problems range from cortical brain signal acquisition and processing to X-Ray image analysis to text and genomics data-mining and subsequent knowledge discovery.