COMPUTER-ASSISTED REGISTRATION, SEGMENTATION, AND 3D RECONSTRUCTION FROM IMAGES OF NEURONAL TISSUE-SECTIONS

Neuroscientists have studied the relationship between nerve cell morph ology and function for over a century. To pursue these studies, they n eed accurate three-dimensional models of nerve cells that facilitate d etailed anatomical measurement and the identification of internal stru ctures. Although serial transmission electron microscopy has been a so urce of such models since the mid 1960s, model reconstruction and anal ysis remain very time consuming. We have developed a new approach to r econstructing and visualizing 3D nerve cell models from serial microsc opy. An interactive system exploits recent computer graphics and compu ter vision techniques to significantly reduce the time required to bui ld such models. The key ingredients of the system are a digital ''blin k comparator'' for section registration, ''snakes,'' or active deforma ble contours, for semiautomated cell segmentation, and voxel-based tec hniques for 3D reconstruction and visualization of complex cell volume s with internal structures.