The University of Sussex

Computer-based neuroradiology training using a 'small worlds' approach to differential diagnosis

Lisa Hinkley, George du Boulay, Catriona Good, Benedict du Boulay, Briony Teather, Derek Teather, Mike Sharples, Nathan Jeffery

The MR (Magnetic Resonance) Tutor is a user-friendly, interactive computer-based training system for neuroradiologists, designed to:(1) teach a structured Image Description Language (IDL) based on anatomical and pathological lesion pattern descriptors and radiological diagnoses, (2) provide a database of cases representing neurological and neurosurgical conditions, (3) reveal intra- and inter-disease variability, and (4) provide a clinically useful decision support training tool using a 'small worlds' diagnostic approach. The database contains blinded formal structured descriptions and all abnormal MR T1, T2, and proton density weighted images of 1200 cases representing a wide spectrum of pathologies. Patterns of contrast enhancement have also been described. A survey of expert neuroradiologists was conducted in order to develop the small worlds to be implemented as the next phase of the system. A list of 21 significant differential diagnoses (small worlds) has been developed. By incorporating cases from the archive which exemplify the small worlds found, the Tutor's training capability will be enhanced. It has the potential to include other digitised neuroradiological imaging techniques, for instance other MRI methods and radio-nuclide imaging. These would require extensions of the existing IDL.


Download PDF