Early diagnosis of dementia based on intersubject whole-brain dissimilarities

Abstract

This article studies the possibility of detecting dementia in an early stage, using nonrigid registration of MR brain scans in combination with dissimilarity-based pattern recognition techniques. Instead of focussing on the shape of a single brain structure, we take into account the shape differences within the entire brain. Imaging data was obtained from a longitudinal, population based study of the elderly. A set of 29 subjects was identified, who were asymptomatic at the time of scanning, but were diagnosed as having dementia within 0.7 to 5 years after the scan, and a set of 29 age and gender matched healthy controls were selected. Each subject was registered to all other subjects, using a nonrigid registration algorithm. Based on statistics of the deformation field in the brain, a dissimilarity measure was calculated between each pair of subjects, yielding a 58×58 dissimilarity matrix. A kNN classifier was trained on the dissimilarity matrix and the performance was tested in a leave-one-out experiment. A classification accuracy of 81% was attained (spec. 83%, sens. 79%). This demonstrates the potential of whole-brain intersubject dissimilarities to aid in early diagnosis of dementia.