Radiological interpretation and diagnosis involves the comparison and class
ification of complex medical images and is typical of the categorisation ta
sks that have been the subject of observational studies in Cognitive Scienc
e. This paper considers the affinity between statistical modelling and theo
ries of categorisation for naturally occurring categories. Statistical base
d measures of similarity and typicality with a probabilistic interpretation
are derived. The utilisation of these measures in the support of diagnosis
under uncertainty via interactive overview plots is described. The applica
tion of the methodology to magnetic resonance imaging of the head is consid
ered. The methods detailed have application to other fields involving archi
ving and retrieving of image data.