Optimisation in Robust Linear Registration of Brain Images

FMRIB Technical Report TR00MJ2
(A related paper has been submitted to Medical Image Analysis)

Mark Jenkinson and Stephen Smith

Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB),
Department of Clinical Neurology, University of Oxford, John Radcliffe Hospital,
Headley Way, Headington, Oxford, UK

Abstract:

Registration is an important component of medical image analysis and for analysing large amounts of data it is desirable to have fully automatic registration methods. Many different automatic registration methods have been proposed to date, and almost all share a common mathematical framework -- one of optimising a cost function. To date little attention has been focused on the optimisation method itself, as opposed to other aspects of the problem like defining suitable cost functions. However, the success of most registration methods hinges on optimisation method. This report examines the assumptions underlying the registration problem and shows that the use of local optimisation methods together with the standard multi-resolution approach is not sufficient to reliably find the global minimum. In addition, a global optimisation method is proposed that is specifically tailored to this registration problem. A full discussion of all the necessary implementation details is included as this forms an important aspect of any practical method. Furthermore, results are presented that show that the proposed method is more reliable at finding the global minimum than several of the currently available registration packages in common usage.

Keywords: affine transformation, global optimisation, multimodal registration, multi-resolution search, robustness