PDF version of this entire document

Experimental Plan for
Assessment of Non-Rigid Registration in 3-D

Abstract:

This collection of notes will be treated as an experimental plan. For context, relevant background is prepended as well. Herein it is assumed that 3-D data, which has been non-rigidly registered in advance, is available for assessment of the quality of registration. Granted, 4 different registrations are put to the test and a distinguishable difference - in terms of quality - exists among them. The document describes steps which need to be taken in order to evaluate and compare these 4 registrations of one 3-D brain dataset. Each registration makes use of a different algorithm, the intent being to conduct a benchmark-type study. In order to ensure that this laborious process is fruitful, it is vital to run evaluations in parallel, assessing the results as they emerge and get gradually refined.