Statistical methods for the analysis of longitudinal brain imaging data
in dementia.
EPSRC funded Industrial CASE PhD studentship with GlaxoSmithKline, in
collaboration with the Dementia Research Centre at the Institute of
Neurology UCL. The research will be done under the auspices of leading
institutions in statistics and medical imaging, with the potential to
generate important advances in the use of imaging for diagnosis and
management of dementia.
For informal discussion contact Professor Mike Kenward
([log in to unmask]) or Chris Frost ([log in to unmask]).
This three year award covers full tuition fees and maintenance for UK
residents and tuition fees only for other EU nationals. Details on
industrial CASE awards can be found on the EPSRC website
(http://www.epsrc.ac.uk).
To apply, send a current CV with the names of two referees, and a
covering letter to
Sarah Thorne, Medical Statistics Unit, London School of Hygiene of
Tropical Medicine, Keppel Street, London WC1E 7HT. The closing date for
applications is 16/07/2004.
The London School of Hygiene and Tropical Medicine is an exempt charity
dedicated to providing excellence in teaching, learning and research for
the benefit of health worldwide.
Further Details:
Dementia presents a growing public health problem in ageing
populations. With the advent of potential therapies, there has been an
increasing demand for tools to aid early diagnosis. In addition,
effective markers of disease progression may play an important role in
assessments of drug efficacy. The aims of this research project are to
develop new statistical procedures for the analysis of voxel level
longitudinal data in structural brain Magnetic Resonance Imaging (MRI)
studies by integrating existing techniques primarily developed for
functional imaging with methods for the statistical analysis of
longitudinal data, and to apply these techniques to improve diagnosis
and understanding of neurodegenerative diseases.
The project will draw together three strands of research. Firstly
existing statistical techniques have been developed for the analysis of
3D brain images at the voxel level. Such analysis attempts to directly
analyse the approximately million signal intensities (arising in an
approximate 100x100x100 3D grid) that make up each scan. There has been
much work in this area; however most of it has been in functional
imaging applications. Functional imaging involves repeated short-term
stimuli being applied to subjects with multiple short-term images
acquired. An aim of this project will be to see to what extent such
techniques can be used when interest focuses on long-term structural
changes.
Secondly automated techniques have been developed at the Dementia
Research Centre for positional matching, or registration, of serial
images on the same subject. Such techniques substantially reduce the
impact of measurement error. Registration of serial MRI can be used to
generate difference images, which consist of many thousands of voxel
level measurements, each representing change over time in an individual.
The correlation structure of these voxel level measures of difference
has strong implications for the development of analytical techniques and
this will be explored in the project.
Thirdly in parallel with the growth in imaging technology over the past
decade there has been substantial growth in statistical methodology that
is relevant to longitudinal studies of brain imaging. Models that can
accommodate the potentially complex pattern of variability within- and
between-images can be formatted and fitted within the framework of
models for hierarchical and dependent data using well established
comuter intensive methods. Such methods have achieved great success in
other areas of the life sciences including biological modelling and
environmental epidemiology.
The project is collaboration between three institutions, the Medical
Statistics Unit at the London School of Hygiene and Tropical Medicine,
the Dementia Research Centre at the Institute of Neurology UCL and
GlaxoSmithKline. The Medical Statistics Unit has major research
interests in the analysis of longitudinal data and both co-supervisors,
Professor Mike Kenward and Chris Frost have worked extensively in this
field. The Dementia Research Centre is one of the leading imaging
centres in the UK. Dr Nick Fox, who will provide a clinical lead, is an
expert in the field of medical imaging and responsible for the
development of the image registration techniques that will be used in
the project. The Dementia Research Centre has a carefully archived
database of clinical and research MRIs. This includes serial imaging of
over 300 subjects with a variety of dementias. A wide range of
between-scan intervals, from same-day to several years, are available
due to prospective imaging of individuals with dementia and controls.
These data represent a powerful resource for the work proposed.
GlaxoSmithKline has a portfolio of on-going studies involving medical
imaging and a commitment to developing new techniques in this area. In
anticipation of the Clinical Imaging Centre (opening in 2006),
GlaxoSmithKline is building up a strong group of imaging scientists, MR
physicists and clinicians to perform cutting-edge research in medical
imaging. This base of knowledge, and resources available at
GlaxoSmithKline, provide a unique compliment to the academic
institutions involved.
Diana Elbourne
Professor of Health Care Evaluation
Medical Statistics Unit
London School of Hygiene and Tropical Medicine
Keppel Street
London WC1E 7HT
UK
tel: +44 (0)20 7927 2629
fax: +44 (0)20 7637 2853
email [log in to unmask]
http://www.lshtm.ac.uk/eph/msu/
and
Professor of Evidence-Informed Policy and Practice,
Social Science Research Unit
Institute of Education
18 Woburn Sq
London WC1H 0NR
Tel: 020 7612 6578
Fax 020 7612 6400
Email [log in to unmask]
http://eppi.ioe.ac.uk/education.htm
|