Visualizing data in three dimensions

In microscope-based image applications, two-dimensional slices viewed in three dimensions allow complex cell geometries to be studied in a more comprehensible manner. But with such displays, it may be difficult to perceive the links between two-dimensional (2-D) data and three-dimensional (3-D) renderings. Graphic contours drawn over 2-D data and assembled in 3-D can help linkages to be seen, but rendering into 3-D surfaces is often necessary to display continuities clearly.

Mar 1st, 1997

Visualizing data in three dimensions

In microscope-based image applications, two-dimensional slices viewed in three dimensions allow complex cell geometries to be studied in a more comprehensible manner. But with such displays, it may be difficult to perceive the links between two-dimensional (2-D) data and three-dimensional (3-D) renderings. Graphic contours drawn over 2-D data and assembled in 3-D can help linkages to be seen, but rendering into 3-D surfaces is often necessary to display continuities clearly.

"Despite this, it is common for errors to be introduced when constructing such models," says James Kremer of the University of Boulder (Boulder, CO). Kremer and his colleagues have developed a software package called IMOD for analyzing and viewing 3-D data from tomographic, serial section, or optical sections. "Because identifying and correcting errors in a rendered model is impossible," says Kremer, "it is necessary to image data in a number of different ways."

IMOD software currently runs on 24-bit graphic work stations from Silicon Graphics (Mountain View, CA). Using the MRC image file format, developed at the Medical Research Council (Cambridge, England), the software allows 3-D image data to be viewed as slices, wireframes, or a voxel projection of 3-D data. Once a 3-D volume has been modeled, its projection can be displayed either by surface or volume rendering. Quantitative results, such as volumes, surface areas, and centroids can then be obtained from model data.

Even with the development of such models, says Kremer, a trained scientific observer is still the best judge of difficult questions about the relationships between 3-D models and original 2-D data. For more information, contact Kremer at emer@beagle. colorado.gifdu.

More in Home