A Digital Atlas of Neurochemical Anatomy in the Hippocampus

$\begin{figure} \setlength {\fboxsep}{0mm} \setlength {\fboxrule}{1pt} \centerline{ \fbox {\epsfig{file=hippo.eps, width=\linewidth}} }\end{figure}$

Introduction
The hippocampus is a part of the temporal lobe, which has a well established role in learning, memory and emotion. This is supported by the finding that certain brain disorders such as schizophrenia and Alzheimer's are associated with temporal lobe pathology. Changes in anatomic morphology and neurochemistry are often reported in these cases.

Despite many studies, the anatomical characteristics of the major neuropsychiatric disorders are still poorly understood. Furthermore, few rapid and sensitive techniques exist for characterising morphological variation of neural structure with which pathology can be identified. Presently, studies depend upon fairly coarse and simplistic measurements such as anatomic volume or thickness, measures which are unable to register anything other than the most gross of structural and neurochemical abnormalities. This may be particularly inappropriate for complex 3D structures such as the hippocampus.

Statistical Models
The purpose of this project is to develop statistical methods for accurately and rapidly characterising normal and abnormal variation of neuroanatomic structure. This is to be achieved by the construction of 3-dimensional atlases of the hippocampus from stained autoradiographic sections of normal, schizophrenic and Alzheimer affected hippocampi.

Statistical Appearance Models are the result of techniques
developed in the Wolfson Unit. The shape and grey-level appearance of an image (or image volume) can be modelled using this paradigm, allowing the encoding of variability in a compact and specific way.

Measurement of Normal and Abnormal Hippocampi
We aim to utilise Statistical Appearance Models to characterize the 3D shape and neurochemical make-up of normal and abnormal hippocampi, resulting in a compact parameter-space in which hippocampi can be described and compared. This will allow the detection of pathology which is more subtle than can be presently identified.

The project has several technical components

1.: Automated registration of 2D serial sections
2.: Automated segmentation of 2D serial sections
3.: Digital 3D reconstruction of data
4.: Construction of 3D statistical appearance models

Detecting Lateralities
Hippocampi are located in both left and right hemipsheres of the brain. Neuroscientists are often interested in identifying differences between structures that are ``paired'' in this way. The form and magnitude of these differences (so called ``lateralities'') are also a factor when identifying pathology due to brain disorders.

A second part of this project is developing tools for accurately and rapidly characterising left-right hemisphere lateralities in normal and abnormal hippocampi. The statistical technique of Linear Discriminant Analysis has been employed to successfully identify structural and neurochemical lateralities in 2D sections of the hippocampus. With completion of the 3D reconstruction process, this technique can be applied to whole 3D hippocampi.

Summary
In summary, the key novel aspects of this approach are

1.: Statistical characterisation of biological and pathological variation
2.: Statistical characterisation of hemispheric lateralites
3.: Use of structural and neurochemical information

Acknowledgements

Funding for this project was provided by the EPSRC and involves collaboration with the department of Psychiatry.

Daniel Poxton

Research Courses Clinical Radiology Industrial liaison Contact Us Join Us Search Home

Top Page