ERCIM News No.29 - April 1997

Techniques for Brain Morphometry

by Alain Colchester and Gérard Subsol

BIOMORPH is a European Community project that began in May 1996. Its aim is to develop improved techniques for measurement of the size and shape of biological structures (morphometry). Morphometry has become increasingly important since the advent of non-invasive magnetic resonance imaging (MRI) which is providing large volumes of data from normal as well as abnormal individuals and allows time changes to be studied.

The techniques to be developed are generic, but the project is focused on schizophrenia and multiple sclerosis, applications in neurology and psychiatry where the need for improved morpho-metric techniques is particularly clear. These are common diseases of great social and medical as well as economic importance. In schizophrenia, changes in the morphology of various brain structures are thought to provide important clues to the most fundamental brain abnormalities that underlie the condition, but the changes are barely detectable with current techniques and more research is needed to confirm and extend existing data. In multiple sclerosis, quantification of changes in lesions has become of great importance in serial MRI studies of how the abnormalities relate to clinical events and for pharmaceutical trials where improved morphometry could reduce the cost of developing new drugs.

The basic requirements for improved morphometry are similar in the different applications. Measurement of size is hampered by the difficulties of detecting and localising the boundaries of structures in images (segmentation), particularly in three dimensions. Time-series analysis is limited by the ability of accurately finding corresponding points in scans taken at different intervals (registration). However, recent advances in both segmentation and registration could now strengthen morphometric studies. Experts regularly describe normal and abnormal shape in the language of anatomy and medicine, but developing and validating computational techniques has proved difficult. In the BIOMORPH project, we will develop and validate improved techniques for measurement of size, shape and changes over time of brain structures shown on MR images in schizophrenia and multiple sclerosis. The clinical research collaborators are active in clinical research and will be able to take immediate advantage of technological advances that arise through the project.

During the first phase of the project MR images will be segmented interactively by the clinical experts and simple morphometry carried out. The anatomical objects thus formed will then be studied by the experts and traditional (mainly linguistic) shape descriptors applied to them. Computer shape descriptors will also be applied and will be added to the data base of objects and their attributes. This sequence will be repeated in the second phase using advanced computer tools developed earlier in the project. Finally, these data will be analysed to establish the differences in size and shape measurements within the group of experts, within normal subjects, within patients suffering from one of the clinical disorders, and between normals and each clinical category. The effectiveness of the computer methods will then be evaluated by comparing them with the expert methods in their reproducibility and their ability to segregate normal from abnormal subjects. The partners of the project are: University of Kent, UK; Project Epidaure, INRIA, France; SGFI-ETH, Switzerland; University of Oxford, UK; Katholieke Universiteit Leuven, Belgium.

Please contact:
Alain Colchester - University of Kent
Tel: +44 1227 827 201

Nicholas Ayache, Gerard Subsol, Jean-Philippe Thirion - INRIA Sophia Antipolis
Tel: +33 4 9365 7668
E-mail: {Nicholas.Ayache, Gerard.Subsol,Jean-Philippe.Thirion}

return to the contents page