Skip to content

Latest commit

 

History

History
344 lines (219 loc) · 8.48 KB

Introduction.md

File metadata and controls

344 lines (219 loc) · 8.48 KB
marp size
true
14580

Data analysis for Neuroimaging - PSYG4043 / C84DAN

Overview

Denis Schluppeck, JeYoung Jung

What's the plan?

  1. Acquire some [functional] MRI data in a simple, but real experiment

  2. Analyze the data with two commonly used software packages spm (matlab) and fsl

  3. Learn a bit about UNIX, organising data and code: version control, in particular git and github

  4. Use different tools to inspect and visualise data

  5. [optional] anatomical, diffusion weighted +/- multi-echo data (T2*)

Learning objectives

  1. Build a better understanding of principles (and practice of) magnetic resonance imaging

  2. Learn about state of the art analysis tools for (f)MRI

  3. Appreciate usefulness of scripting, unix, version control for reproducibility / transparency of work

  4. Discover matlab, toolboxes for data visualisation

Timeline

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

What's the assignment?

A short, written report

Summarise the experimental setup, analysis methodology and results. Need to have clearly written abstract (250w), methods, results and discussions (and figures).

Aim: Get you thinking about journal-style writing, rather than essays. Plus: presenting your own data, identifying key points, a story/pitch.

When to work on this? 👷

Start as soon as we have the data

  • explore your analysis ideas
  • talk to us about questions you could address
  • think about plots + data visualisations you'd like to make

Submission details 📎 📚

Currently w/ Student Services, date to-be-confirmed

  • turn-it-in submission on moodle page
  • deadline: end of March (the week after last class of this module - check moodle page for details)

What's the assignment (2) 📊 📈 📉

  • 250w abstract
  • plus a main document (max 1500w)
  • references / citations as for standard written work
  • max 5 figures1 illustrating
    • details of the experimental setup
    • analysis methodology
    • results
1figures can have sub-panels or subplots

❗ For next time (lab 2)

  • sign up for 1 of 3 groups (max 7 people) - moodle
  • complete visitor screening form
  • we also need a volunteer (~40 min in scanner)

Setting up computers, logins

Let's check log-ins and make sure we can find:

  • Terminal
  • Matlab & set up paths for SPM

Note

Each user (at a particular machine) needs to make sure that Terminal/shell is set up correctly – if you move machines between labs (or want to do things elsewhere, repeat the above step!)

Set up colors in terminal

alias ls='ls -G'

# or append to bash_profile
echo "alias ls='ls -G'" >> ~/.bash_profile

in Terminal ✔️

Has setup worked? fsl

Look at some existing anatomies with fsleyes ✔️

which fsl # see anything?

fsleyes & # File -> Add Standard -> Pick 1st or 2nd

do you see

SPM/FSL analysis

  • get data from sessions S001to S004 into a common folder data
  • make folders, copy files by "drag & drop"
  • point & click version (like some of you have already done)
  • digging into the details of how this is implemented
  • inspecting analysis output, intermediate files, ...
cd ~/data/S001/  # for example
# run SPM analysis in matlab (JJ)
# run FSL analysis (DS)

Some UNIX

  • only basics needed for running fsl
  • lots of functionality is available through point-and-click
  • but command line is helpful for organising data!
  • more complex analysis, e.g. freesurfer, require some working knowledge
# navigate file system
# cd, ls, pwd, which, ...

# some powerful commands for organising your data
# cp, rm, touch, mkdir, rmdir

# some stuff to show of how powerful
# grep, "lists", "wildcards (*, ., ?)"
# "regular expressions"

Version control git

  • 30min lecture on principles of version control (git)
  • start using your (free) github.com id by working on a simple project
  • make your first modifications to a local copy of code and get it into a repo.
mkdir test && cd test # what does this do?
git init
# [[ create, edit a file, say my_first.md ]]
git add my_first.md  # add it to "staging area"
git commit # enter commit message
# - OR -
git commit -m 'adds first version of file'
git log

Version control (v2.0) 😉

Everyone should sign up for a free github account, so we can work together on this from session 4 onwards: https://github.com/join

  • it's free and useful
  • we'll want to play with this in lab #4
  • once you have an username (pick one that I will recognise!), go to our github classroom ... details to follow

matlab - reading images (1)

  • we'll learn/revisirt how to read imaging data into matlab (nifti files)
% > R2017b
data = niftiread('file_from_scanner.nii');

matlab - reading images (2)

  • revisit indexing of arrays, "slicing", etc.
data(12, 24, :, 1) % what is this?
data(:, 24, 24, 1) % ... and this?
data(34, 44, 12, :) % ... or that?
  • we built a returnSlice() function, to complete imageviewer:
% function signature
s = returnSlice(array, sliceNum, orientation);
  • some more coding along these lines

makeMontage('dafni_01_anatomy.nii', 25)

function [  ] = makeMontage(fname, nSamples)
%makeMontage - make a montage from 3d/4d image

matlab - timeseries and subplots

Using toolboxes to visualise data (JJ)

  • glass brains
  • SPMs, cluster maps on anatomical images.
  • cut-aways, ...

Wrap-up (Lab 7)

  • recap what have we covered in the last 7 weeks?
  • where to go to from here (unleash your inner coding 🐯)
  • try to approach each new problem, project where you find lots of repetition (analysis, writing, coding, ...):
    • there must be a better way!
    • what's the smallest unit that gets repeated all the time?
    • can I use bash/unix, matlab or another tool to automate?
  • just try things out - you'll learn tons in the process

For next time

  • make sure you complete screening form (and/or get in touch with any concerns)

  • to volunteer for a scan, send me an e-mail: [email protected]

  • next week: see you at SPMIC for your timeslot - if you don't know where on campus, google SPMIC... first hit is imaging centre w/ directions

Random technical stuff

Some notes of manipulating text files in the shell (using awk) and (viz timing files, etc.)

Notes - text hacking in shell

Small awk program for adding a counter n and time t and turn one column txt file into csv file:

awk 'NF {print NR-1 ", " (NR-1)*1.5 ", "  $1}' \
     timecourse.txt > timecourse.csv
  • with a headerline (matlab's csvread() doesn't like!)
awk 'BEGIN {print "n, t, response"}
     NF    {print NR-1 ", " (NR-1)*1.5 ", "  $1}' \
     timecourse.txt > timecourse.csv

Solution in matlab?

Turn timecourse.txt (column of y-values), into timecourse.csv:

  • where first column is a counter that goes from 1...n,
  • the second column is t (in s), which goes up from 0..1.5s.. and
  • the third column is the y values

What about something else?

  • Excel? R? Another cool idea that's worth having in your set of tools?