Biomedical Engineering Reference
In-Depth Information
anatomy, and genetics. As part of this, scientists from various domains collab-
oratively worked on data analysis and synthesis while drawing from an enor-
mous amount of multidisciplinary data available at various scales and levels
of resolution. Case study results are derived from user experiences reported
by neuroscientists, clinical physicians, statisticians, and computer scientists. The
evaluation of these reports confi rms that the proposed visualization and ana-
lytics framework is an effi cient mechanism to detect and validate expected
information but most importantly an instrument aiding with the discovery of
unexpected information contained within the multiscale multimodality data.
Within this context, it had to be possible to process and fuse multimodality
and multiscale biomedical data at interactive rates at the genomic, proteomic,
cellular synaptic, psychometric, and behavioral levels, requiring the develop-
ment of a range of processing, visualization, and interaction techniques. Of
particular importance were techniques for the representation of conventional
structural and functional two- and three-dimensional (2D, 3D) imaging data
as well as brain activation patterns, derived information such as brain tractog-
raphy, genetics information in the form of results from SNP array runs con-
ducted for each patient, and digitally published reference material including
archival publications via PubMed and elsewhere. Once this was possible,
domain experts were in the position to concurrently and collaboratively work
toward extracting new insights while also signifi cantly shortening data process-
ing times when compared against traditional methods, which in one particular
case led to a discovery of genetic markers for schizophrenia, previously con-
sidered years of work away.
27.3.1
Visualization Practices in Imaging Genetics
Scientifi c visualization is pervasive in many biomedical research areas.
Visualization tools usually utilize 3D or 2D algorithms to render a single data
modality such as computerized tomography (CT), magnetic resonance imaging
(MRI), or microscopic data. Examples are volume rendering and isosurface
rendering on a set of radiology data. Users are commonly domain experts that
understand the particular data and thus can change the rendering parameters
to visually interpret the region of interest and highlight important aspects to
others. Further reasoning can be based on the visual data or from the actual
raw data. One such example is the Visualization Toolkit (VTK) [41]. VTK
provides a set of implementations of common visualization algorithms and
programming interfaces are defi ned for customizing applications. Some other
tools go one step further by integrating image processing algorithms. One such
example is the Insight Segmentation and Registration Toolkit (ITK) [42]. This
toolkit was designed to support the Visible Human Project and has become a
platform for fundamental segmentation and registration algorithms. Another
example is 3D Slicer [43], which is based on VTK and ITK to support visual-
ization and image analysis capabilities for biomedical data. In bioinformatics,
the enormous amount of data make it feasible to combine visualization and
Search WWH ::
Custom Search