Image Processing Reference
In-Depth Information
23.3.10 Population Imaging
In population imaging, medical image data and other measurements are acquired of
a large group of subjects, typically more than one thousand, over a longer period,
typically years, in order to study the onset and progression of disease, general aging
effects, and so forth in larger groups of people. Examples include the RotterdamScan
Study focusing on neuro-degeneration [
46
] and the Study of Health In Pomerania
(SHIP) focusing on general health [
35
].
This application domain is an extreme example of multi-subject medical visual-
ization discussed in Sect.
23.2
, integrating large quantities of heterogeneous, multi-
modal and multi-timepoint data acquired of a large group of subjects. The scientists
running these studies usually do not formulate strictly-defined hypotheses before-
hand, instead opting for meticulous data acquisition, followed by an extended period
of analysis in order to extract patterns and hypotheses from the data. Recently, Steen-
wijk et al. [
63
] set the first steps for the visualization of population imaging by
applying visual analysis techniques to cohort study imaging data. The extreme het-
erogeneity and magnitude of the data, coupled with the explorative nature of the
research, renders this a promising long-term application domain for visual analysis
and medical visualization.
23.4 Conclusions
In this chapter, we gave a compact overview of the history of medical visualization
research, spanning the past 30 years. Based on this history and on our own observa-
tions working in the field, we then identified and discussed the research challenges
of the coming decade.
Our discussion of classic medical visualization problems related to efficient and
high quality display of
one
static dataset was brief.We devotedmore space to data that
change over time, to the integration of anatomy with simulation and finally to cohort
studies. We refer to problems where such time-dependent and high-dimensional data
are employed as
MedVis 2.0
problems. While the classic problems are—from an
application perspective—solved, there are many research opportunities in MedVis
2.0 problems. These data are significantly more difficult to analyze, to process and to
visualize. Time-dependent MRI data, e.g., exhibit all artifacts of static MRI data but
a number of additional artifacts, e.g. due to motion. Integrated analysis and visualiza-
tion is a key feature of MedVis 2.0 solutions. In general, successful solutions to these
problems require a considerably deeper understanding of the medical background
and thus favor close collaborations with medical doctors over merely having access
to medical data.
Search WWH ::
Custom Search