Accurate and precise 3D surface based registration

Abstract

Image registration is an important task in image-guided pre-surgery planning, surgery and post-surgery analysis. Fudicial markers are commonly used to achieve accurate registration. This technique provides a near real time, robust and accurate registration. However, due to it's invasive nature, it is not as popular as the group of noninvasive techniques that make use of intrinsic features for registration. The main challenge in using intrinsic features is that the accuracy in extracting these features directly affects the registration accuracy. Surface-based registration is one of the most commonly used technique for structural registration (head registration, bone registration, etc). It does not require the use of external markers and is suitable for cross modality registration, computed tomography (CT) and magnetic resonance (MR) images. However, this technique requires the precise extraction of the corresponding surfaces and also an optimization routine to compute the optimal transformation relating the reference and current surface. In this paper, we will describe a near automatic procedure for accurately extracting bone surfaces from CT image. The experimental results show that the registration attained is in the order of sub-pixel accuracy.