Non-rigid registration using higher-order mutual information

D. Rueckert, M. J. Clarkson, D. L.G. Hill, D. J. Hawkes

Research output: Contribution to journalConference articlepeer-review

148 Scopus citations

Abstract

Non-rigid registration of multi-modality images is an important tool for assessing temporal and structural changes between images. For rigid registration, voxel similarity measures like mutual information have been shown to align images from different modalities accurately and robustly. For non-rigid registration, mutual information can be sensitive to local variations of intensity which in MR images may be caused by RF inhomogeneity. The reason for the sensitivity of mutual information towards intensity variations stems from the fact that mutual information ignores any spatial information. In this paper we propose an extension of the mutual information framework which incorporates spatial information about higher-order image structure into the registration process and has the potential to improve the accuracy and robustness of non-rigid registration in the presence of intensity variations. We have applied the non-rigid registration algorithm to a number of simulated MR brain images of a digital phantom which have been degraded by a simulated intensity shading and a known deformation. In addition, we have applied the algorithm for the non-rigid registration of eight pre- and post-operative brain MR images which were acquired with an interventional MR scanner and therefore have substantial intensity shading due to RF field inhomogeneities. In all cases the second-order estimate of mutual information leads to robust and accurate registration.

Original languageEnglish
Pages (from-to)I/-
JournalProceedings of SPIE - The International Society for Optical Engineering
Volume3979
StatePublished - 2000
Externally publishedYes
EventMedical Imaging 2000: Image Processing - San Diego, CA, USA
Duration: 14 Feb 200017 Feb 2000

Fingerprint

Dive into the research topics of 'Non-rigid registration using higher-order mutual information'. Together they form a unique fingerprint.

Cite this