Overview of the WAIR 2.0 Software
The WAIR software is a tool for quantitative analysis of various n-dimensional
(n-D) image registration techniques. In particular, its applications include,
but are not limited to, analyzing warp performance for stereotactic
Human Brain anatomical and functional data. Using the Discrete Wavelet
Transform (WT) we develop several image registration (warp) classification
schemes. The first one, called "wavelet-space triangle analysis", is
applicable for studying a family of warps on a single or multiple
n-D data sets. For each data set the
WAIR routine assigns a positive real number to every warp alignment
in the family, and the best warp, for the given data,
will be the one having the smallest
value associated with it. The second classification method, called
"cluster group classification", is applicable for analyzing the overall
performance of a family of warps of a groups of data sets. Here, there is
a single number assigned to each registration alignment, based on its
group-clustering characteristics. Third goodness of warp approach, called
SGC (spread group classification), is applicable for analyzing functional
brain data. It gives preference to registration techniques that spread
apart baseline versus activation functional signal for group data.
Both, the multi-dimensional discrete WT and the corresponding "triangle",
"cluster-group" and "spread-group" classification approaches are included
in the package.
All routines are invoked as standard UNIX
commands and (with one exception)
neither expect nor allow user interaction once the command has been
issued. The series of 'C' subroutines which comprise the WAIR library can
be easily incorporated into the user's site specific programs adapted to their
A future version of WAIR will include a graphical user interface (GUI)
written in Tcl/Tk.
Currently, the WAIR package does not provide any graphics or image display
capabilities. And no file format converters are included in this software.
The user is supposed to convert their original data into 4 byte floating-point
raw data files, then execute the WAIR routines to obtain the desired warp
This package was originally developed to work on Sun SPARC stations using the
'C' language compiler
provided by Sun as part of the standard system software.
If you currently use an 8 bits/pixel
file format on a Sun SPARC station equipped with the standard 'C' compiler,
have at least (XX Bytes + 16 MB)
of RAM (where XX = 4*(Number of data sets)*(Number of warps)*(Dimension_1 * ...
*Dimension_n) Bytes), and are familiar with the UNIX operating system,
you should be able to install and use the WAIR
package without additional assistance, even if you know nothing about
'C' programming. However, if any of
these conditions is not met, it is likely that you will need the assistance of
a 'C' programmer who is familiar with
the UNIX operating system.
There are several 2D test images provided in the subdirectory
"test_data.dir/" that can be used to inspect the installation
of the software, as well as, for testing and interpreting the results of
the wavelet analysis, both analytically and visually.
© 1997 Ivo D. Dinov,
( > )