go-v.orgdoi

Excursion sets of random fields with applications to human brain mapping

Jin. Cao · container not stated · 1997

  • DOI 10.82308/2014
  • Publisher OpenAlex
  • Type dissertation
  • Issued 1997

Not cached yet. This record was fetched live for this request and is stored after the lookup. Reopening the page should serve the cached copy.

· OpenAlex · dissertation · 1997

Abstract

The statistical analysis of brain functional and structural change presents a formidable statistical problem. Computation of a test statistic indicating change at each location of the brain gives a 3-D statistical image. The global maximum of the statistical image has been proposed by Friston et al. (1991) and Worsley et al. (1992) to test for regions of high change based on the belief that the changes are localized, not distributed. To find the correct p value, Worsley et al. (1992) modelled the statistical image as a stationary random field under the null hypothesis.

The approximate p value of the global maximum was then determined by the expected Euler characteristic (EC) of the excursion set, that is, the set of points with a value larger than a given threshold. The expected EC of excursion sets of stationary Gaussian random fields has been derived by Adler and Hasofer (1976) and Adler (1981) and extended to $ chi sp2, t$ and F fields by worsley (1994). The first part of the thesis aims to extend their results to Hotelling's $T sp2$ fields and apply the theory to a real example from 3-D brain deformations.

Although the global maximum as a test statistic is more widely used, some workers have suggested an alternative: the size of the largest connected component of the excursion set, based on the belief that neural activity tends to stimulate changes over contiguous locations. Nosko (1969) derived the asymptotic distribution of the size of one connected component of the excursion set above high levels for a stationary Gaussian field. Applying this work, Friston et al. (1994) presented an approximate analysis for assessing the significance of focal changes by their spatial extent.

Although the analysis was developed for Gaussian fields originally, it was suggested that it could also be applied to some non-Gaussian fields, like the t field, by appropriate "Gaussianization", that is, pointwise transformation to a Gaussian distribution. However, the Gaussianzied field is not a Gaussian random field and hence the approach is not strictly valid. The second part of the thesis aims to derive the asymptotic distribution of the size of one connected component of the excursion set for $ chi sp2, t$ and F fields and apply this to an experiment on pain perception.

Open the resolved locator · 10.82308/2014

Metadata licence: OpenAlex metadata (CC0)

Registry
OpenAlex DOI index + API
Fetched
2026-11-20 03:23 UTC
Cache state
not cached yet (bulk-index+openalex)
Locator
https://hdl.handle.net/20.500.14905/96026
Metadata licence
OpenAlex metadata (CC0)
Scope
Metadata as returned by the resolver; abstracts are never generated here.