Interactive Cluster
Analysis ToolKit
The Interactive Cluster Analysis Toolkit has been designed
to be an environment for use in the identification of clusters.
Clusters, also known as connected components in the field of
image analysis, can most simply be described as a collection of pixels
gathered together under a template or neighborhood rule.
ICAT comes in two parts. The analysis tool itself is written in C++.
It utilizes the Enhanced Hoshen-Kopelman algorithm to provide the
method for cluster identification. A graphical user interface (gui),
written in Java, accompanies this, providing an interface which allows
a more intuitive use of the tool. While the gui may currently only
be used on machines which have direct network access, the C++ aspect of
ICAT may be run independently.
This software was designed initially to attempt
to provide one method for the analysis of images from diabetic retinopathy,
the study of the human eye when afflicted with diabetes. While the
initial intentions were medical in nature, this tool is general-purpose;
it will perform an analysis on any image provided in the correct format,
giving statistical results based upon the metrics currently in place.
Disclaimer: The GUI (Graphical User Interface)
for ICAT was initially developed using JDK 1.0 with some components
later migrated to JDK 1.1. The authors do not guarantee
compatiblity (for the GUI) across all computing platforms.
- Publications
- Graphic User Interface (GUI)
demonstration
- How to get a copy of ICAT
- The People
ICAT / berry@eecs.utk.edu