Distributed multiuser visualization of time varying anatomical data

We describe a networked environment for navigating and visualizing 3-dimensional anatomical data with extensions for time varying volumes. The Duke Center for In Vivo Microscopy (CIVM) has been capturing volumetric data of mice using magnetic resonance microscopy. Current data sets are 512/sup 3/ with 16 bit precision per voxel at an isotropic resolution of 50 microns. A new instrument will provide larger 512*512*2048 volumes. Because magnetic resonance imaging is nondestructive, both rapid time series and longer interval developmental series can be taken from living specimens. Our work builds on techniques put in place at the Pittsburgh Supercomputing Center and the University of Michigan for navigating Visible Human data using a client-server implementation, but applied to CIVM mouse data. Extension of the system to 4-dimensional data sets involves changes to compressed data representations and client viewing mechanisms. An essential aspect of the mouse studies is to facilitate comparison between different specimens, or even the same specimen over time, for studies of morphologic phenotype expression in gene knockouts.