Towards automated electrocardiac map interpretation: an intelligent contouring tool based on Spatial Aggregation
by
L. Ironi, S. Tentoni
Proc. of 5th International Symposium on Intelligent Data Analysis, Berlino, 28-30 Agosto, 2003, Lecture Notes in Computer Science, 2810, 397-408.
ABSTRACT
The interpretation of cardiac potential maps
is essential for the localization of the anomalous excitation sites associated with electrical conduction pathologies, such as arrythmia, and its automation would make clinical practice
both realistic and of great impact on health care. Conventional contouring
tools allow us to efficiently visualize patterns of electrical
potential distribution but they do not facilitate the automated extraction
of general rules necessary to infer the correlation between pathopysiological patterns
and wavefront structure and propagation. The Spatial Aggregation
(SA) approach, which aims at interpreting a numeric
input field, is potentially capable of capturing structural information
about the underlying physical phenomenon, and of identifying its global patterns
and the causal relations between events. These characteristics are due to
its hierarchical strategy in aggregating objects at higher and higher levels. However, when dealing with a complex domain geometry and/or with
non uniform data meshes, as in our context, the original version
of SA may unsoundly perform contouring.
Isocurve entanglement and/or segmentation phenomena may occur due to metric-based adjacency relations and a scarse density of isopoints. This paper discusses the problems above, and presents definitions and algorithms for a sound representation of the spatial contiguity between isopoints and for the construction of isocurves.
Download Postscript file.
Back to publication list
Liliana Ironi
2003