A PhD studentship funded by the National Medical Laser Centre at UCL is
available for a project to study classification using high-dimensional
Funding: The award covers home fees and a maintenance stipend of at least £10,500
per annum (free of tax) in 2004/5.
Eligibility: This studentship is open to all applicants. Non-EU applicants
would need to be able to fund the additional cost of overseas fees.
Applications to Kathryn Dempsey, PA to Professor Bown, at the National Medical
Laser Centre, 67-73 Riding House Street, London W1W 7EJ, tel 020-7679-9060,
email [log in to unmask]
Closing date: 20th September
Researchers at the National Medical Laser Centre, UCL, are studying the use of
photodynamic therapy (PDT) as a means of preventing a condition called Barrett's
oesophagus developing into cancer of the gullet. To remove abnormal cells in the
oesophagus the treatment employs a drug that makes the cells light sensitive and a
laser, delivered via an endoscope, that kills them.
The project for which the studentship is available arises from the possibility of using
light as a complementary tool for diagnosis, monitoring and treatment. A lamp
delivers light spanning a broad range of wavelengths, and there is evidence that
normal and abnormal cells have different patterns of scattering and absorbance. It
should be possible to use measured scattering and absorbance spectra to classify areas
of cells as normal or abnormal, or even grade the level of abnormality. Data are
available, and more can be collected, that can be used to train a classification rule.
This is a challenging statistical problem, because the spectral data have very high
dimensionality, there may be as many as 1000 wavelengths in a measured spectrum,
and the spectral differences between normal and abnormal cells are responsible for
only a small part of the variability in the spectra.
There are already encouraging results using this technique (elastic scattering, ESS) for
detecting precancerous and early cancerous changes in the oesophagus, colon, breast,
skin and mouth, but much remains to turn this approach into a technique useful for
routine clinical use. The idea of ESS for monitoring PDT is entirely new.
There are several possible approaches to the problem, many of which involve
reducing the dimension of the spectral data in some way, for example by principal
components analysis. There is also plenty of scope for developing new approaches or
extending and improving existing ones. The aim of the project is twofold: to find an
approach that works well for this problem, and to try to make some methodological
contributions to the area of classification with high dimensional data as a whole.
The principal supervisor would be Professor Tom Fearn, in the Department of
Statistical Science, but the student would be based working with the research team at
the National Medical Laser Centre. This is led by Professor Stephen Bown and Dr
Laurence Lovat and includes physicists, clinicians and laboratory scientists.