Philipp Muigg, Vienna, AT
from 15:00 to 16:30
|Where||R 309, Albert-Einstein-Str.21|
|Add event to calendar||
Direct Time Series Visualization Techniques in the Context of Medical Perfusion Data
This talk will focus on a novel visualization technique applicable to large time series data. Different rendering and interaction methods will be demonstrated by applying them on medical perfusion data. The presented visualization techniques are based on an "image based" clustering algorithm which allows for interactivity even for large datasets containing hundred or thousands or even millions of data items. Different post processing effects, such as edge enhancement, line integral convolution, or a novel coloring scheme are presented in order to improve different aspects of the time series visualization.
The interaction techniques presented in this talk allow for the fast and intuitive selection of time series. This is achieved by letting the user sketch interesting patterns and calculating a distance measure based on the time series's similarity to this pattern. We will present two application examples dealing with medical perfusion data from ischemic stroke and breast cancer diagnosis.