[Comp-neuro] Interactive Visualization of Multiresolution Image Stacks in 3D

[Shawn Mikula]

The following article is now available at
http://brainmaps.org/index.php?p=publications

The paper deals with rapid interactive visualization of very high-resolution 
neuroanatomical data, and hopefully will encourage and facilitate the 
development of neuroanatomically inspired models.

Trotts I, Mikula S, and Jones EG

Interactive Visualization of Multiresolution Image Stacks in 3D

ABSTRACT
Conventional microscopy, electron microscopy, and imaging techniques such as 
MRI and PET commonly generate large stacks of images of the sectioned brain. 
In other domains, such as neurophysiology, variables such as space or time 
are also varied along a stack axis.   Digital image sizes have been 
progressively increasing and in virtual microscopy, it is now common to work 
with individual image sizes that are several hundred megapixels and several 
gigabytes in size.   The interactive visualization of these high-resolution, 
multiresolution images in 2D has been addressed previously (Sullivan & 
Baker, 1994).  Here, we describe a method for interactive visualization of 
multiresolution image stacks in 3D.

The method, characterized as quad-tree based multiresolution image stack 
interactive visualization using a texel projection based criterion, relies 
on accessing and projecting image tiles from multiresolution image stacks in 
such a way that, from the observer's perspective, image tiles all appear 
approximately the same size even though they are accessed from different 
tiers within the images comprising the stack.  This method enables efficient 
navigation of high-resolution image stacks.  We implement this method in a 
program called StackVis, which is a Windows-based, interactive 3D 
multiresolution image stack visualization system written in C++ and using 
OpenGL. It is freely available at http://brainmaps.org.