N2G.Spatial.Mixture: fMRI Spatial Mixture Modelling

In AnalyzeFMRI: Functions for Analysis of fMRI Datasets Stored in the ANALYZE or NIFTI Format

Description

Fits the spatial mixture model of Hartvig and Jensen (2000)

Usage

N2G.Spatial.Mixture(data, par.start = c(4, 2, 4, 2, 0.9, 0.05),
                    ksize, ktype = c("2D", "3D"), mask = NULL)

Arguments

data	The dataset (usually a vector)
par.start	Starting values for N2G model
ksize	Kernel size (see paper)
ktype	Format of kernel "2D" or "3D"
mask	Mask for dataset.

Value

p.map = a1, par = fit$par, lims = fit$lims Returns a list with following components

p.map	Posterior Probability Map of activation
par	Fitted parameters of the underlying N2G model
lims	Normal component interval for fitted model

Author(s)

J. L. Marchini

References

Hartvig and Jensen (2000) Spatial Mixture Modelling of fMRI Data

See Also

N2G.Class.Probability, N2G.Likelihood.Ratio, N2G.Density , N2G.Likelihood , N2G.Transform, N2G.Fit , N2G , N2G.Inverse , N2G.Region

Examples

## simulate image
d <- c(100, 100, 1)
y <- array(0, dim = d)
m <- y
m[, , ] <- 1

z.init <- 2 * m
z.init[20:40, 20:40, 1] <- 1
z.init[50:70, 50:70, 1] <- 3

y[z.init == 1] <- -rgamma(sum(z.init == 1), 4, 1)
y[z.init == 2] <- rnorm(sum(z.init == 2))
y[z.init == 3] <- rgamma(sum(z.init == 3), 4, 1)

mask <- 1 * (y < 1000)

## fit spatial mixture model
ans <- N2G.Spatial.Mixture(y, par.start = c(4, 2, 4, 2, 0.9, 0.05),
                           ksize = 3, ktype = "2D", mask = m) 

## plot original image, standard mixture model estimate and spatial mixture
## model estimate

par(mfrow = c(1, 3))
image(y[, , 1])
image(y[, , 1] > ans$lims[1]) ## this line plots the results of a Non-Spatial Mixture Model
image(ans$p.map[, , 1] > 0.5) ## this line plots the results of the Spatial Mixture Model

AnalyzeFMRI documentation built on Oct. 5, 2021, 5:06 p.m.