A review study comparing conventional intensity-based and geometry-based registration methodologies.
Three major registration frameworks were examined: (a) intensity-based, exhaustive registration framework using three distinct cost functions (b) geometry-based registration framework, featuring three geometrical descriptors (c) the original implementation of the Iterative Closest Point algorithm.
Use of geometrical feature descriptors for aligning 3D medical data.
All compared techniques were applied to CT data pairs with known and unknown initial spatial differences.
Geometry-based and intensity-based techniques perform similarly, as far as accuracy is concerned, but geometry-based methods significantly reduce processing time.