Medical Informatics :
           Modelling Uncertainty in Neuroradiological Diagnosis

                    Department of Medical Statistics
             Faculty of Computing Sciences and Engineering
                         De Montfort University
                                 Leicester

                              EPSRC PhD Studentship

Applications are invited for an EPSRC Research Studentship to work with an 
established multi-institution multi-disciplinary team working in the area 
of computer support for medical image interpretation, training and diagnosis 
in neuroradiology. 
The studentship pays EC fees and provides support for 3 years.

Start date : Available immediately or by negotiation.

Applicants should have a good first degree and be able to demonstrate a 
sound knowledge of statistics and relevant computing skills.

                          Background to the Studentship

The proposed research work will build on an existing long term collaboration 
between the Department of Medical Statistics; the  Institute of Neurology, 
London; the School of Cognitive and Computing Sciences, University of 
Sussex; and the Educational Technology Research Group, University of 
Birmingham. The Medical Systems Research Group, De  Montfort University, 
in collaboration with the Institute of  Neurology has developed a defined 
description language for  describing abnormalities visible on Magnetic 
Resonance images of  the head, an archive of some 1200 cases with confirmed 
diagnoses  and a statistical database of case descriptions. The image and  
statistical archive form a valuable resource which is  fundamental to the 
project. A recently completed  multidisciplinary study has resulted in the 
development of novel  overview spaces which provide a visualisation of the 
variation  in image appearance within pathologies and statistically based  
measures of the cognitive science concepts of typicality and  similarity. 
The proposed research will be informed by an ongoing  joint ESRC funded 
project into 'Workbased Studies in Radiology'.  The collaboration has 
resulted in the development of a prototype  Intelligent Tutoring System 
for MR imaging of the Head. 

Project Description                

Radiological diagnosis and decision making presents a   particularly 
difficult task in that it involves the extraction and interpretation 
of information on individual patients from complex medical images and 
reasoning in relation to competing pathologies that often exhibit 
similar abnormal features in the    images. The detail available in 
Magnetic Resonance (MR) images,    the sensitivity of image appearance 
to sequence type, the variability of presentation makes diagnosis of 
brain diseases    particularly challenging. Significant progress in 
support of    radiological interpretation is only likely to be made if 
an integrated approach is adopted which allows for the uncertainty    
and imprecision of the tasks involved and allocates tasks    between 
the computer and clinician that make maximal use of the expertise. 

Radiologists are rapidly able to detect abnormalities visible in images 
but may have only a partial understanding of the    relationship between 
the appearance of brain lesions and the nature of brain disease and are 
poor at characterising  the    uncertainty in the diagnostic process. 
This project will undertake fundamental research into the provision of 
computer based representations to support radiologists in understanding 
and visualising uncertainty and imprecision in image    interpretation 
and diagnosis. The work will build on recent research into the provision 
of 'overview spaces' of pathologies.  In particular, the issues to be 
addressed include methods for visualising the uncertainty in the 
description of abnormalities, the provision of well-calibrated estimates 
of    diagnostic uncertainty and the visualisation of this uncertainty in 
overview displays for competing diagnoses. 

For further information please contact Dr Briony Teather in the
Department of Medical Statistics email : [log in to unmask], Tel 0116 2577480.
******************************************************************************
Dr Briony A Teather
Principal Lecturer in Medical Statistics
Executive Conference and Short Course Secretary
Department of Medical Statistics	          Tel : +44 (0)116 2577480
De Montfort University				  Fax : +44 (0)116 2506114
The Gateway					  Email : [log in to unmask]
Leicester LE1 9BH
******************************************************************************




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%