Image Processing Reference
In-Depth Information
information about shapes not only from realistic shading but also from appropriate
hatching and from outlines that support the mental separation of nearby objects
rendered in similar colours.
Illustrative visualization is related to the term Non-Photorealistic Rendering in
computer graphics, or NPR for short. The termNPRwas used since around 1990when
the seminal paper of Saito et al. clearly illustrated that complex 3D shapes could be
renderedmore comprehensible by using certain feature lines [ 61 ]. Compared to NPR,
illustrative visualization is the more focused term that covers rendering techniques
serving clear visualization goals, namely to convey shape information efficiently.
In medical visualization, either surfaces or volume data are rendered in illustrative
styles. For illustrative volume rendering, the term volume illustration was introduced
by Ebert et al. in 2000 [ 20 ]. Boundary enhancement based on gradient approxima-
tion [ 17 ] and curvature-based transfer functions [ 40 ] are landmarks in illustrative
medical visualization. Tietjen et al. applied silhouettes and other feature lines for
various scenarios in liver surgery planning [ 68 ]. Besides silhouettes, stippling and,
probably even more, hatching, have great potential to reveal details of shapes [ 32 ].
Later, Bruckner et al. [ 10 - 12 ] made a number of important contributions that
support depth and shape perceptionwith adapted transfer functions. In particular, they
considered the peculiarities of interactive exploration of 3D datasets and elaborated
on the idea of preserving essential context information. These and later refinements
are integrated in theVolumeShop-system that is publicly available and used by several
research groups.
23.2.6 Multi-subject Data
Medical visualization has also started to work on the problem of dealing with multi-
subject data. These are datasets that include measurements, including imaging, of
more than one subject. The goal is to be able to extract patterns that affect sub-
groups of the whole collection, for example to explore which aspects of the data
correlate with a specific disease outcome. An example of this type of work includes
LifeLines2, an information visualization approach to visualize and compare mul-
tiple patient histories or electronic medical records [ 73 ]. More recently, work has
been done on the interactive visualization of the multi-subject and mixed modality
datasets acquired by medical cohort studies [ 63 ]. In these studies, mixed modality
data, including imaging, genetics, blood measurements and so on, are acquired from
a group of subjects in order to understand, diagnose or predict the clinical outcome
of that group. Steenwijk et al. demonstrated that it was possible to create a highly
interactive coupled view visualization interface, integrating both information and
scientific visualization techniques, with which patterns, and also hypotheses, could
be extracted from the whole data collection.
 
Search WWH ::




Custom Search