Sørensen, Helle4; Goldsmith, Jeff2; Sangalli, Laura M3
1 Department of Mathematical Sciences, Faculty of Science, Københavns Universitet2 Columbia University3 Politectico di Milano4 Department of Mathematical Sciences, Faculty of Science, Københavns Universitet
Functional data are data that can be represented by suitable functions, such as curves (potentially multi-dimensional) or surfaces. This paper gives an introduction to some basic but important techniques for the analysis of such data, and we apply the techniques to two datasets from biomedicine. One dataset is about white matter structures in the brain in multiple sclerosis patients; the other dataset is about three-dimensional vascular geometries collected for the study of cerebral aneurysms. The techniques described are smoothing, alignment, principal component analysis, and regression.
Statistics in Medicine, 2013, Vol 32, Issue 30, p. 5222-5240