REVIEW OF INTENSITY-BASED AND GEOMETRY-BASED REGISTRATION TECHNIQUES, APPLIED ON 3D CONE-BEAM CT DATA

4^th International Conference on Medical Informatics & Telehealth

October 6-7, 2016 | London, UK

George K. Matsopoulos

National Technical University of Athens, Greece

Posters & Accepted Abstracts: J Health Med Informat

Abstract :

In numerous medical applications, precise spatial alignment of Computed Tomography (CT) images is a strong requirement. Alignment is achieved by applying digital image processing techniques known as image registration. Typically, registration considers the intensities or color of all data points (image pixels), thus dramatically increasing processing time, especially in large data sets, such as CT volumes. Alternative approaches consider only a selected set of key-points from the examined data sets. Those key-points can be fully described using various geometrical characteristics, instead of their color information, thus forming unique geometrical descriptors. The geometrical descriptors can be used for allocating corresponding points between two compared data sets, which in turn can be used for aligning them. This study focuses on the comparison of several geometry-based, descriptor-oriented registration techniques, as well as conventional, exhaustive, intensity-based methods for aligning three-dimensional (3D) Cone-beam CT (CBCT) data pairs. Specifically, three general image registration frameworks were examined: (a) a geometry-based methodology featuring three distinct geometrical descriptors, (b) an intensity-based methodology using three different similarity metrics and (c) the Iterative Closest Point algorithm. All methodologies and their derivatives were applied for a total of thirty 3D CBCT data pairs with both known and unknown initial spatial differences. The results were assessed both qualitatively and quantitatively and it was concluded that the featured geometry-based registration framework performed similarly to the examined exhaustive registration techniques, by achieving significantly improved processing time.

Biography :

Email: gmatso@esd.ece.ntua.gr