Biomedical Engineering Reference
In-Depth Information
Figure 27.13
Overview and data analysis modalities synthesized within the
workspace.
Another interactive discovery case involved a dichotomy of data and aimed
at the detection of brain functional signifi cance. In this case the analysis
process starts with a collection of fMRI images with associated underlying
characteristics such as clinical test results. Image data are then clustered based
on the activation on anterior lobe versus posterior lobe, that is, the motor
associated part against the sensor perceptive part. The result is then further
clustered by the dorsal versus ventral half of the brain and refi nement can
continue until something signifi cant is identifi ed or termination criteria met.
For the presented system, this simple dichotomy clustering is implemented as
a drag-and-drop operation, which has shown to compress the process of mul-
tiple selection rounds to hours, a fraction of the time originally needed to sift
through thousands of printouts. The latter is a process that can take days or
weeks to complete, and researchers may be pressed to recall what they saw in
a particular print-out viewed the day before, rendering the traditional discov-
ery process ineffi cient. These types of analysis methodologies are exemplary
for the targeted imaging genetics research.
27.5
CONCLUSIONS
A driving force that brought researchers initially together was the curiosity
for an entirely different collaborative visual analytics space, which quickly was
adopted as a productivity tool as part of weekly meetings. With the size of data
sets growing, researchers found that they were able to sort and categorize their
data faster and more accurately and in turn identify trends and subjects with
distinctive characteristics. At the same time, the team-centric analysis pass
allowed them to identify areas where data should be analyzed more thor-
oughly or additional data points would be desirable. The team identifi ed fi ve
Search WWH ::
Custom Search