Integrated Brain Image Modelling (IBIM)

Overview

A collaborative project performed by Imaging Sciences at the University of Manchester together with The overall aim of the project was to produce a system for analysing structural brain images, which will aid segmentation of brain substructures and quantification of how they relate to each other in relative size, position and shape.

Aims

The following aims summarise the Manchester contribution to the project goals

Approach

Fundamental to achieving the above aims is the need to find dense correspondences across a set of brain images. Recent work at Manchester by Tim Cootes, Carol Twinning and Stephen Marsland (now at University of Auckland) has investigated obtaining such correspondences using non-linear diffeomorphic warps. This built on previous work by Rhodri Davies on using Minimum Description Length (MDL) as a framework for creating groupwise objective functions for measuring the quality of correspondences.

Built on the framework for constructing automatic statistical models of shape and appearance using groupwise non-rigid registration as proposed by Cootes et. al. [1]. We will construct models of sub-structures in a set of brain images in which these sub-structures have been labelled. The models will be used to experiment with segmentation of the sub-structures in a leave-one-out manner. Results will be compared with currently available alternatives.

Results

So far we have applied a groupwise registration algorithm to establish correspondences between images such that we do not have to explicitly select one member of the group as a reference for pairwise registration. We have used the correspondences obtained to build 3D active shape models (ASMs) and 3D active appearance models (AAMs). We are now using these models to search unseen images and automatically segment out subcortical structures.

Publications

Comparing the Similarity of Statistical Shape Models using the Bhattacharya Metric. KO Babalola, TF Cootes, B Patenaude, A Rao, and M Jenkinson. In Proceedings of the 9th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), Lecture Notes in Computer Science, vol. 4190, pp. 142-150, October 2006

Cannonical Correlation Analysis of Sub-cortical Brain Structures Using Non-rigid Registration. A Rao, Kolawole Babalola, and Daniel Rueckert. In Proceedings of the 3rd International Workshop on Biomedical Image Registration (WBIR), Lecture Notes in Computer Science, vol. 4057, pp. 66-74, July 2006

Groupwise Registration of Richly Labelled Images. KO Babalola, and TF Cootes. In Proceedings of the 10th Annual Conference on Medical Image Understanding and Analysis (MIUA), pages 241-145. Manchester, UK, July 2006

Canonical Correlation Analysis of Sub-cortical Brain Structures Using Non-rigid Registration. Anil Rao, Kola Babalola, and Daniel Rueckert. In Proceedings of the 10th Annual Conference on Medical Image Understanding and Analysis pages 86-90. Manchester, UK, July 2006

Building 3D Statistical Shape Models using Groupwise Registration. KO Babalola, TF Cootes, CF Twining, V Petrovic, R Schestowitz and CJ Taylor. In Proceedings of the 12th Annual Meeting of the Organisation for Human Brain Mapping, Florence, Italy, June 2006

Registering Richly Labelled 3D Images. KO Babalola and TF Cootes. In Proceedings of the International Symposium on Biomedical Images, Washington, Virginia, USA, April 2006

The following publications give a background to our methods

[1] T.F. Cootes, C.J. Twining and C.J. Taylor, "Diffeomorphic Statistical Shape Models", Proc. British Machine Vision Conference 2004, Vol.1, pp.447-456

[2] T.F. Cootes, S. Marsland, C.J. Twining, K. Smith and C.J. Taylor, "Groupwise Diffeomorphic Non-rigid Registration for Automatic Model Building", Proc ECCV2004, pp.316-327

[3] T.F. Cootes, D. Cooper, C.J. Taylor and J. Graham, "Active Shape Models - Their Training and Application." Computer Vision and Image Understanding. Vol. 61, No. 1, Jan. 1995, pp. 38-59