Statistical Methods for Population Neuroimaging

Lead Research Organisation: University of Warwick
Department Name: Statistics

Abstract

The dissertation topic is in applied statistics in the area of statistical inference for neuroimaging data, a topic that lies between EPSRC's themes of Engineering, Healthcare Technologies, Information and Communication Technologies and of course Mathematical Sciences. The project will develop inference methods that account for the spatial structure of the images, making inference on topological sets instead of individual spatial elements.

This project will develop tools for a newly emerging type of brain imaging study, where instead of collecting data on 10's of subjects, data is acquired on 1000's of subjects. In addition to the brain imaging data, a constellation of other health, behavior and demographic data are also collected. The first priority for this project is to develop inference methods suitable for such large N studies. Presently, traditional null hypothesis testing approaches struggling to detect effects, at most finding (say) 5% of the brain with changes; with 10,000 subjects, such method will find essentially 100% of the brain contains significant changes. The student will develop spatial confidence set methods that produce 3D versions of confidence intervals that will allow users to identify regions showing changes that are practically significant. The subsequent project will address how to integrate the high-dimensional non-imaging data into this inference framework, e.g. developing parametrised spatial confidence sets that are defined as functions of latent subject factors.

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/N509796/1 01/10/2016 30/09/2021
1791128 Studentship EP/N509796/1 03/10/2016 30/09/2017 Alexander David Bowring