Contact Us Intranet


News Home
Video on Demand
Subscribe to Our Newsletter
Frequently Asked Questions

3D volume reconstruction technique improves accuracy and performance

Story posted December 6, 2006

Visualization of reconstructed 3D volume representing vasculogenic mimicry patterns and tumor blood vessels.
Visualization of reconstructed 3D volume representing vasculogenic mimicry patterns and tumor blood vessels

Determining differences in the three-dimensional organization of vasculogenic mimicry patterns and tumor blood vessels may provide important information that can enhance the treatment of a variety of solid cancers. Both vasculogenic mimicry patterns and blood vessels distribute fluid to many highly invasive cancers. To examine these competing patterns, pathologists cut tissue samples into cross-sections a few micrometers thick. They scan each sample several times with a specialized microscope, then reconstruct the scans into a 3D version of the area of interest and visualize the 3D volume reconstructions.

Traditionally, this 3D volume reconstruction is done manually and the visual inspection often suffers from intensity variations in microscopy images. The reconstruction process is time-consuming and prone to error. A recent study published in the Journal of Microscopy, however, shows that automated 3D volume reconstruction software provided significant improvements in accuracy, consistency, and performance time when compared to reconstructions built by human technicians. For example, the automatic process showed alignment errors that were more than 10 times smaller than the manual process.

Another study in the Journal of Microscopy proposed a method for correcting the intensity of microscopic images produced in the volume reconstruction process.

This automated approach to 3D volume reconstruction and the intensity correction technique were developed by NCSA's Peter Bajcsy and Sang-Chul Lee and Robert Folberg and Amy Lin of the University of Illinois at Chicago's College of Medicine.

For further information, see