Practical, computer-aided registration of multiple, three-dimensional, magnetic-resonance observations of the human brain.

摘要

Researchers working with MR observations generally agree that far more information is available in a volume (3D) observation than is considered for diagnosis. The key to the new alignment method is in basing it on available information on surfaces. Using the skin surface is effective a robust algorithm can reliably extract this surface from almost any scan of the head, and a human operator's exquisite sensitivity to facial features is allows him to manually align skin surfaces with precision. Following the definitions, we report on a preliminary experiment where we align three MR observations taken during a single MR examination, each weighting arterial, venous, and tissue features. When accurately aligned, a neurosurgeon can use these features as anatomical landmarks for planning and executing interventional procedures.