Skip to content

Latest commit

 

History

History
129 lines (82 loc) · 8.29 KB

gettingData.md

File metadata and controls

129 lines (82 loc) · 8.29 KB

Getting Data

What's the plan?

Unit Topic
1 Introduction, Administrivia, computers, ...
2 ⭐ Data acquisition (scanning on 3T)
3 Inspecting & analysing data in FSL
4 Version control (git and github.com)
5 Images in Matlab, display, analyze
6 Timeseries signals in Matlab
7 Reading/writing text, CSV, data files Matlab

Data acquisition in the scanner

fMRI data will be acquired in 45min sessions (in small groups) on one of our 3T scanners. Have a look at the webpage for the 3T Philips Achieva to learn a bit more about the machine we will be using.

Two important sets of things to consider:

  • what are the parameters / settings on the MRI scanner?
  • what is the subject doing in the scanner (stimuli & task)?

Scanner - Protocol

The protocol will be pretty standard for a cognitive neuroscience scanning sessions. The plan for the time in the scanner is as follows

  1. Quick survey scan to allow "planning" on console (< 10s)
  2. "inplane anatomy" - anatomical scan with planned slice prescription / coverage (~2min)
  3. fMRI experiment (block): gradient-echo EPI, TR 1.5s, TE 40ms, FA: 72º (1x or 2x ~4min)
  4. fMRI experiment (event-related): same scanning parameters, (1x or 2x ~4min)
  5. T1w-MPRAGE: to illustrate detailed (1mm isotropic) anatomy (~5min)
  6. [sometimes included] EPI data (test): 5 echoes at different TEs to illustrate T2s decay
  7. [sometimes included] field map: to show that B0 can be measured (< 1min)

You can have a look at the Matlab function flipAnglePlots() to remind yourself about why we picked the particular flip angle.

Scanner - actual numbers from the day (2019/20).

For each of three subjects (subject-A, subject-B and subject-C) we collected 5 scans. The "series" numbers may be slightly different across individuals. but the scans (in order) were the same:

  1. fMRI data - 2D gradient-echo EPI, TR 2.0s, TE 30ms, multiband factor 2, FA: 80º, scene_localiser() scan: 192 dynamics. (Check the dimensions in the actual data with fslinfo or fslhd.)

  2. fMRI data - acquisition details as in 1, but only 120 dynamics. Pseudorandomly ordered blocks on "normal", "inverted" and "caricatured" faces (at different levels).

  3. same as 2 (a repeat)

  4. whole head anatomy - T1w 3D MPRAGE. Matrix size 256 x 256, 160 slices. Reconstructed as sagittal images. voxel size 1 mm isotropic. TE 3.7 ms, TR 8.13 ms, FA 8°, TI 960 ms, and linear phase encoding order.

  5. T2 weighted anatomy scan. TE 89ms, TR 3.38s, FA, 90º (Check the dimensions in the actual data with fslinfo or fslhd.)

Scanner - actual numbers from the day (2018/19).

  1. inplane anatomy - T1w, 2D MPRAGE. 24 slices, inplane voxel size: 1.5 mm, slice thickness 3 mm (same prescription as fMRI data). Matrix size 128 x 128.
  2. fMRI data (and all fMRI scans) 2D gradient-echo EPI, TR 1.5s, TE 40ms, FA: 72º, scene localiser scan: 294 (288+4) dynamics, retinotopy scan: 160 dynamics. Voxel size 3mm isotropic (acquired). Matrix size 64 x 64. Note: the scanner reconstruction may have changed the "reconstructed voxel size" to something else, for mathematical expediency. (Check the dimensions in the actual data with fslinfo or fslhd.)
  3. whole head anatomy - T1w 3D MPRAGE. Matrix size 256 x 256, 160 slices. Reconstructed as sagittal images. voxel size 1 mm isotropic. TE 3.7 ms, TR 8.13 ms, FA 8°, TI 960 ms, and linear phase encoding order.
  4. across the subjects we also acquired various other data sets (including diffusion weighted, a T2 weighted anatomy scan (using multiple echoes), ... but we didn't find time to analyze these in class.

Scanner - actual numbers from the day (2017/18)

  1. inplane anatomy - T1w, 2D MPRAGE. 24 slices, inplane voxel size: 1.5 mm, slice thickness 3 mm (same prescription as fMRI data). Matrix size 128 x 128.
  2. fMRI data (and all fMRI scans) 2D gradient-echo EPI, TR 1.5s, TE 40ms, FA: 72º, 160 dynamics. Voxel size 3mm isotropic. Matrix size 64 x 64.
  3. whole head anatomy - T1w 3D MPRAGE. Matrix size 256 x 256, 160 slices. Reconstructed as sagittal images. voxel size 1 mm isotropic. TE 3.7 ms, TR 8.13 ms, FA 8°, TI 960 ms, and linear phase encoding order.

Stimulus set-up

Stimulus code (2017/18)

  • stimulus code: inspect FFAlocaliser and jazLocaliser for lots of details. You should be able to run the code, too.
  • timing: block design 12s rest, 12s images (faces, objects), so 24s cycles. 10 repeats per scan - 240s or 160 dynamics (@ 1.5s). Stimuli were images from face and object image database (details below).
  • timing: event-related design 1.5s stimuli followed be a randomly chosen ITI between 9s and and 18s (in 1.5s increments). Stimuli were white, moving, random dots on a black background. 100% coherence, speed 4 deg/s. Scans lasted 240s or 160 dynamics (@ 1.5s).

The code is written in matlab/mgl using the task library that comes with mgl. Written by Alex Beckett and DS based on a version of a working code from Justin Gardner 😄

There are a couple of short youtube videos explaining the FFA localiser and the fixation dimming task to control attention. This should give you a sense of what the subject is doing inside the scanner.

The experiment runs as a simple block design in the following order:

[faces, rest] , [objects, rest] - ...

The length of each [stimulus, rest] cycle is determined by the cycleLength (in TRs).

To run, make sure the stimulusCode folder is on the path and then simply run the following command. the Escape key can be used to stop the experiment at any point:

FFAlocaliser % quick test to see what's going on

To run at the MR centre, we also want to specify TR, not to run in a small window, etc. So probably worth setting a few parameters in the call like this:

FFAlocaliser('TR=1.5', 'debug=0', 'numBlocks=10', 'cycleLength=12')

Notes

  • working code with face and object images
  • parameters to set cycle length, TR, number of block
  • youtube clip explaining the fixation task and experiment
  • write out text file in correct format for fsl/feat analysis.
  • test with actual scanning parameters

Materials for stimuli:

We will provide the stimulus code (written in Matlab / MGL) in line with what happened in - Learning Matlab / C84NIM - a pre-requisite for this course.

We'll run a "Faces versus houses / scenes localiser, as this works well and is a very robust experiment.

Stimulus images courtesy of Michael J. Tarr, Center for the Neural Basis of Cognition and Department of Psychology, Carnegie Mellon University, https://www.tarrlab.org/. Funding provided by NSF award 0339122.

Object stimuli from: Brady, T. F., Konkle, T., Alvarez, G. A. and Oliva, A. (2008). Visual long-term memory has a massive storage capacity for object details. Proceedings of the National Academy of Sciences, USA, 105 (38), 14325-14329.

Scene stimuli from: Konkle, T., Brady, T. F., Alvarez, G.A. and Oliva, A. (2010). Scene memory is more detailed than you think: the role of categories in visual long-term memory. Psychological Science, 21(11), 1551-1556.