Image Processing Reference
In-Depth Information
Particle systems
are readily applied to blood flow. Commonly, particles are
depicted as spheres [
48
], or otherwise represented by small integral curves [
37
]. Both
approaches convey blood flow speed through color, while direction information is
captured by temporal cohesion of the animated particles. Integral curves addition-
ally provide a short history of the particle trajectory. Both conventional approaches
employ the available visual cues, such as color and shape, to capture only the blood
flow velocity information. Van Pelt et al. [
47
] propose an illustrative particle-based
approach that captures the velocity information by means of shape, keeping the color
cue available for more elaborate blood flow characteristics. They mimic techniques
often used to conveymotion in comic books by deforming a ball at high-speedmotion,
and adding speed lines to improve the perception of direction (see Fig.
25.9
c).
Visual clutter remains an important issue in 4D blood flow field visualizations.
Usually, this clutter is avoided by user biased interaction methods which can miss
important properties of the flow. Grouping vector field areas with meaningful similar
characteristics, i.e., clustering, can help in developing techniques to improve the
visual exploration and minimize the user bias. Some work exists in the clustering of
static 3D vector fields [
10
,
21
,
24
,
43
] while little research has been conducted to
extend it to 3D unsteady flow fields [
55
]. Developing and extending these techniques
to blood flow data is an interesting research direction. The main challenge is to
provide a clustering that has a meaning for the user, and an adequate visualization
technique that enables efficient exploration of the clusters.
25.5 Discussion and Open Issues
Simulated and measured blood flow data have been two distinct research fields that
have developed in parallel. Simulation data is based on many assumptions, it is
difficult to make it patient-specific, and also validation is a challenge. Measured flow
data, on the other hand, represents the patient-specific flow, but it has a lot of limitation
concerning resolution, artifacts, and noise in the data. An interesting direction is to
combine both methods to strengthen each other. For example, measured data can be
used as boundary conditions for a simulation, or simulation methods could be used
to compensate for the lack of temporal resolution.
Blood flow data sets are considerably large data sets since they consist on a
time series of vector-field volumes. The issue of dealing with large data will be of
increasing importance given the improvements in spatial and/or temporal resolution
that are expected.
Recently, new blood flow visualization techniques have been developed. Many
decisions with respect to seeding, segmentation, integral curves, the use of illustration
techniques are rather ad-hoc decisions based on intuition. It is important to link the
decision to the users' needs. Additionally a more thorough exploration of the design
space and comparisons of existing methods is needed. A major challenge is that
this data is new to the domain experts, and it is challenging for them to identify the
relevant features to visualize.
Search WWH ::
Custom Search