Abstract #1801
Evidence of non-normal distributions in brain imaging data from normal subjects: implications for diagnosis of disease
David Alexander Dickie 1 , Dominic E Job 1 , Joanna M Wardlaw 1 , David H Laidlaw 2 , and Mark E Bastin 1
1
The University of Edinburgh, Edinburgh,
United Kingdom,
2
Brown
University, Providence, RI, United States
The most commonly used statistical methods in brain
imaging are parametric, i.e., assume data follow the
normal (Gaussian) distribution. Whether or not
structural brain MRI data follow the Gaussian
distribution, and whether this actually matters, has yet
to be determined. This work tested whether brain MRI
volumes in a typically sized adult sample (n=80; 25-64
years) were Gaussian distributed. The impact of
distribution shape on effect sizes between age groups
was then determined. We found that these data were not
Gaussian. This led to large, unsystematic errors in
parametric effect sizes of normal ageing brain volumes.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
