Visualization in Biology and Medicine - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 241

of the preprocessing step. A large variety of metadata can be derived from the
raw data, the inclusion of which depends on the specific application and scientific
question.
Visualization is necessary in the first three preprocessing steps for data validation
and quality assessment. Most automatic techniques come with errors and uncertain-
ties which need to be conveyed to the user to help them judge the quality of their
data and improve existing algorithms. Computation of metadata is commonly an
important part of the visualization process itself and will be discussed in more detail
in Sect. 22.4.3 .
22.4.3 Visualization and Data Analysis in Evo-devo
A large number of evo-devo frameworks and techniques are being developed to
support comprehensive analysis tasks. Most efforts currently concentrate on data
preprocessing such as feature segmentation and the subsequent mathematical analy-
sis where statistics are a widely applied tool [ 49 ]. Visualization is often reduced to
the depiction of the data, which in itself is already rather difficult due to the size of the
data and its time-dependent and multivariate nature. Tools widely used on the appli-
cation side have a similar feature spectrum. Many of them support image processing
tasks such as data registration and segmentation, feature extraction and analysis, and
interactive rendering of three-dimensional data—even time-series data is supported
by most applications. Examples of such tools are: ( commercial ) MetaMorph, Imaris,
Volocity, Amira; and ( open-source ) ImageJ, Fiji, BioImageXD, V3D.
As the needs of specific applications diversify, specialized tools are being
developed such as Cellenger (automated image segmentation and feature analy-
sis: http://tiny.cc/rARky/ ) , the CellProfiler project (automated image segmentation,
feature analysis, data mining and visualization: http://cellprofiler.org ) , and vari-
ous commercial products from microscope manufacturers. Recent research has also
concentrated on the interactive validation of preprocessing steps applied to 3D spatio-
temporal data. Examples for the validation of segmentation results are a combination
of different rendering modes to overlay raw with segmented data [ 51 ] and the visu-
alization of automatically quantified uncertainty [ 15 , 20 , 37 ]. A second direction is
feature tracking, which has long been an important area of research in visualization
[ 34 ], and is a crucial task in biological data analysis. Two major distinctions can be
made for tracking algorithms operating on volumetric time-series data. One direc-
tion of research focuses on digital image processing approaches such as optical flow
and operates directly on the scalar data [ 12 , 30 ]. The other direction is based on
tracking structures that have been segmented beforehand [ 25 , 27 ]. If individual cells
are tracked over time, they form cell lineages, i.e., binary trees that encode patterns
of cell migration and division. Several browsers for the investigation of cell lineage
data have recently been proposed [ 5 , 32 , 55 ], which combine volume rendering of
the scalar data and graph drawing for the depiction of the lineage tree.