imaging-02

Aim

The learning objectives for this unit are:

• become more comfortable with manipulating 3d and 4d data ("multidimensional arrays")
• solidify our understanding of what it means for something to be 3 (or 4) dimensional
• load in data from nifti files
• get practice writing writing functions in matlab
• get comfortable with functions that have 1 or more input arguments, and 1 or more output arguments

Loading data from nifti files

data = niftiread('/Volumes/practicals/ds1/2019-msc-matlab/example.nii');

Look at the documentation for niftiread!

A function for returning a single voxel timecourse

Write a function returnTimecourse() that returns a 1D timecourse from a 4D dataset (3d space, 1d time).

Users should be able to use the function like this

c = [24,25,10]; for example
tc = returnTimecourse(data, c);
plot(tc)

• What are the things that you might want to guard against when writing your code?
• Can you anticipate some potential user errors?
• Additional: how can you find out what the TR is to create a vector of time in seconds (rather than units of volumes)?

A function for calculating the HRF

The function makeHrf() creates the y-values for an HRF. The time runs from 0 to 30s (in steps of the TR you specify). So for TR=1.5 it would be [0, 1.5 3, 4.5, ... 30]

t = 0:1.5:30; %
h = makeHrf(1.5); % creates an hrf for 30s
plot(t, h ,'ro-')

Bonus: where to find the equations that describe HRF as a function to time? Are there several available? Could you add a flag to your code that could allow you to switch between different versions?

Also...

Toward the end of this unit, we need to construct a function makeDesignMatrix() that will help us do linear regression in due course. The idea here is to specify the on/off timing (say, for a visual stimulus in an fMRI experiment)

• 2 obligatory input arguments: blockOne, n
• 2 optional input argument: TR, hrf
• returns:
  • d (1-column design matrix taking into account the HRF)
  • dRaw (same as d, but without HRF).

Checking results

Look at help makeDesignMatrix to get some ideas on how to plot. With some additional adjusting you can end up with something like this:

Notes

• the file makeDesignMatrix.p is the pcode/compiled version that runs (but for which you cannot see the code). The .m file here is just to provide the help text - you need to add the code to implement the function yourself!

• dRaw contains n repeats of blockOne in succession,

• d is the same as dRaw, but convolved with a haemodynamic response function.

• d should be returned such that mean(d) = 0

• ninja skills: how to handle non-blocked designs?