79 lines
2.1 KiB
R
79 lines
2.1 KiB
R
# Lab 11 for the University of Tulsa's CS-6643 Bioinformatics Course
|
|
# Introduction to fMRI Analysis and ICA
|
|
# Professor: Dr. McKinney, Fall 2022
|
|
# Noah L. Schrick - 1492657
|
|
|
|
## Set Working Directory to file directory - RStudio approach
|
|
setwd(dirname(rstudioapi::getActiveDocumentContext()$path))
|
|
|
|
#### Part A: Haemodynamic response functions (HRF) and block design
|
|
## Plot Basic HRF from 0-20s
|
|
hq <- function(t,q=4){
|
|
# q=4 or 5, where 5 has more of a delay
|
|
return (t^q * exp(-t)/(q^q * exp(-q)))
|
|
}
|
|
|
|
# use seq to create vector time and use hq to create hrf vectors
|
|
time <- seq(0,20)
|
|
hrf1 <- hq(time)
|
|
hrf2 <- hq(time, 5)
|
|
|
|
# plot
|
|
plot(time,hrf1,type="l")
|
|
lines(time,hrf2,col="red")
|
|
|
|
## Deconvolve with task onset times
|
|
# grabbed from afni c code
|
|
# basis_block_hrf4 from 3dDeconvolve.c
|
|
HRF <- function(t, d){
|
|
if (t<0){
|
|
y=0.0
|
|
}else{
|
|
y = 1/256*exp(4-t)*(-24-24*t-12*t^2-4*t^3-t^4 + exp(min(d,t))*(24+24*(t-min(d,t)) + 12*(t-min(d,t))^2+4*(t-min(d,t))^3+(t-min(d,t))^4))
|
|
}
|
|
return(y)
|
|
}
|
|
|
|
t=seq(0,360,len=360)
|
|
onsets=c(14,174,254)
|
|
blocks.model = double()
|
|
for (curr_t in t){
|
|
summed_hrf=0.0
|
|
for (start in onsets){
|
|
summed_hrf=summed_hrf+HRF(curr_t-start,20)
|
|
}
|
|
blocks.model = c(blocks.model,summed_hrf)
|
|
}
|
|
|
|
plot(blocks.model,type="l")
|
|
|
|
|
|
## Plot voxel time series data and the block design curve
|
|
voxel.data <- read.delim("059_069_025.1D")
|
|
plot(seq(1,2*dim(voxel.data)[1],by=2),t(voxel.data),
|
|
type="l", xlab="time",ylab="intensity")
|
|
# normalize the height of the blocks model
|
|
blocks.normal <- max(voxel.data)*blocks.model/max(blocks.model)
|
|
lines(blocks.normal,type="l",col="red")
|
|
|
|
# regression
|
|
length(blocks.normal) # too long
|
|
dim(voxel.data)[1]
|
|
# grab elements from blocks.normal to make a vector same as data
|
|
blocks.norm.subset <- blocks.normal[seq(1,length(blocks.normal),len=dim(voxel.data)[1])]
|
|
length(blocks.norm.subset)
|
|
|
|
voxel.data.vec <- matrix(unlist(voxel.data),ncol=1)
|
|
|
|
# use lm to create voxel.fit <- lm...
|
|
voxel.fit <- lm(voxel.data.vec ~ blocks.norm.subset)
|
|
|
|
|
|
plot(blocks.norm.subset,voxel.data.vec,
|
|
xlab="block model",ylab="voxel data",main="regression fit")
|
|
|
|
# use abline(voxel.fit) to overlay a line with fit coefficients
|
|
abline(voxel.fit)
|
|
|
|
|