Visuelle, computer-gestützte Medizin

Visual computing addresses various aspects of the processing of image data. These aspects include the full visual computing pipeline starting from data acquisition, via image processing, the graphical representation of the data, to the interaction with them. These aspects have been in the focus of my work of the past 13 years, and became the focus of the research group "Visual Computing for Medicine" (VCM). Although we concentrate on the medical domain as major application fields, I would like to stress that the research contributions are addressing general problems of image processing and computer graphics. Furthermore, they can be applied to many other application fields. The following five parts structure the content into major stages of the visual computing pipeline, while the chapters focus on the specific contributions to the pipeline stages, mostly in the context of medical application. Part I starts discussing the fundamentals of medical imaging in Chapter 2. Specifically, it gives an overview on the structure of volumetric image datasets in Section 2.1, and describes typical data acquisition modalities, such as Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and several more (Section 2.2). Since volumetric data is of a discrete nature, the fundamentals of signal theory - the sampling theorem (Section 2.3.1) - and the source of image artifacts will be discussed in Section 2.3. The consequences of incorrect sampling for discrete volumetric image data are widespread and lead to typical artifacts like aliasing (Section 2.3.2), the partial volume effect (Section 2.3.3), interpolation artifacts (Section 2.3.4), and signal artifacts (Section 2.3.5) itself. The second part discusses approaches on the enhancement and filtering of volumetric image data in Chapter 3. Namely, it addresses the necessary windowing operation, where a higher dynamic range - or high precision voxel value range - is mapped into a smaller one. In order to maintain a good contrast, an operator for high dynamic range windowing is introduced and its effectiveness for volumetric image data is demonstrated.Visuelle, computer-gestützte Medizi