Analysis of Brain Tissue Spatial Correlation by Means of Variography

Magnetic resonance imaging is an excellent tool to non-invasively obtain images from the human body with high soft-tissue contrast. In brain imaging one can easily tune the contrast between white matter, grey matter, cerebro-spinal fluid (CSF) and deep brain structures. However, it is difficult to obtain information about the integrity of the structure of a given tissue type from visual inspection. Nevertheless, normal aging processes as well as neuro-degenerative diseases, such as Alzheimer dementia for example, seem to be accompanied by structural changes of the white matter tissue [e.g. Bartzokis et al., Arch Neurol. 2003].

The present project aims at quantifying spatial homogeneity and correlation and also at identifying the structural anisotropy of brain tissue at different length scales. The basis of the analysis is formed by a common geostatistical tool called variography. These methods will be developed and tailored to the special requirements of brain tissue analysis of MR images.

A further aim of the project is to provide an accurate description of brain tissue "normality" with a few parameters only, which are relatively easy to calculate and can be obtained from a simple 3D scan. This can become a crucial tool for the evaluation of studies involving a substantial cross-section of the population, and have an impact on the evolution of personalised medicine.

The variogram, gamma (colour coded), is calculated for six different directions within a region-of-interest of a single white matter slice. This provides information about the anisotropy of spatial homogeneity and corresponding correlation distances of the sample.