# fMRI Lab for 23th Feb 2023

###### With me (Will), Christina and George

---

### Aim: To get acquainted with the tools needed to generate fMRI results

---

# Plan
#### 1. Tools
#### 2. Introduce experiment
#### 3. View the raw data
#### 4. Explore the design file
#### 5. Preprocessing
#### 6. Modelling task events
#### 7. View the results
#### 8. Multi-subject analysis and automation

---


----

### Tools for neuroimaging


###### FSL (FMRIB Software Library) is a free suite of applications from Oxford's Functional Magnetic Resonance Imaging of the Brain (FMRIB) laboratory.

###### http://mriquestions.com/best-fmri-software.html

----


###### FEAT (FLAME), BET, FSLeyes

###### https://fsl.fmrib.ox.ac.uk/fsl/fslwiki

---

# The Experiment

Before looking at the data we will describe the experiment.

### Aim: To find the neural correlates of word-generation

###### Data and more info: https://www.fmrib.ox.ac.uk/primers/intro_primer/ExBox11/IntroBox11.html

----

This dataset is from an event-related language experiment and has three different types of events:

----

###### - Word-generation events (WG): Here the subject is presented with a noun, say for example "car" and his/her task is to come up with a pertinent verb (for example "drive") and then "think that word in his/her head". The subject was explicitly instructed never to say or even mouth a word to prevent movement artefacts.
###### - Word-shadowing events (WS): Here the subject is presented with a verb and is instructed to simply "think that word in his/her head".
###### - Null-events (N): These are events where nothing happens, i.e. the cross-hair remains on the screen and no word is presented. The purpose of these "events" is to supply a baseline against which the other two event types can be compared.

----

##### Within one session, the events were presented at a constant ISI (Inter Stimulus Interval) of 6 seconds. For example, the first 72 seconds (twelve events) in this session may have looked like:

### N-WS-N-WS-N-WS-N-WG-N-WS-WG-N

---

## The Data

----

# Structural MRI

#### Structural images provide information about the anatomy of what's being imaged by showing differences between different kinds of tissues (Grey Matter vs White Matter vs CSF)

#### Structural MRI provides high spatial resolution (0.8mm x 0.8mm x 0.8mm) images.

---

# Voxels


----

## Over to you (Viewing a structural image)
1. Open (type) fsleyes
2. File -> Add from file -> fsl_course_data/ExBox11/structural.nii.gz
3. Increase brightness + contrast

----


##### _Voxels_: Volumetric Pixel - the basic unit of measurement. Each voxel has a coordinate in the volume

##### _Slice_: all voxels in 2D plane.

##### _Volume_: The entire 3-D grid covering the space imaged in the MRI scanner. Volume is composed of voxels

##### _TR_: One volume scanned every TR - basic time measurement in MRI

---

# What about the 'f' in fMRI?

### We want to map _function_ to _structure_.

### But how do we quantify neuronal activation using non-invasive MRI?

---

## BOLD fMRI!
### _BOLD_ = Blood Oxygenation Level Dependent

#### We measure the level of blood oxygenation throughout different cortical regions, and infer neuronal activation from that.

##### Based on the premise that increased blood oxygenation = increased neuronal activation.

---

# Time

### Each voxel has an associated time-series of BOLD activation.

---



###### We have measurment of BOLD for each voxel across time

---

## The raw fMRI data


- Let's view the raw 4D data for the subject

----

#### - `home/fsluser/fsl_course_data/ExBox11`

- Navigate to this directory, using the `Terminal` (or using the `files` GUI)
- use `cd your_path` to change working directory
- use `ls` to list files in working directory
- keep this terminal window open for later

----

1. Open program 'FSLEyes'
###### - Type `fsleyes` in Terminal/Command Line and press enter
###### - Or, open it from the FSL GUI

2. Add functional data
###### - `file` -> `add from file` -> Find: `/fsl_course_data/ExBox11/fmri.nii.gz`

3. View in Movie Mode

---

## The Analysis

- To get from this raw data to the results, an analysis was conducted using `FEAT`.

- We will look at the details of this analysis now.

----

## Analysis with FEAT

#### 1. Open FEAT GUI
###### - Type `fsl` in terminal & `Enter` -> Click FEAT button
#### 2. Load the design file (`.fsf`)
###### - `Load` -> `/fsl_course_data/ExBox11/fmri.feat/design.fsf`

----

##### You can think of the design file as a recipe that turns your ingredients (raw Nifti data) into a full meal (results). Having this recipe saved is useful because it means you can fully reproduce the meal again.

###### Because this dataset has already been analyzed with FEAT, a design file already exists - but for your experiments, you will need to create one yourself using this software.

##### So, let's explore the design file tab by tab.

----

## `Data`


----

## `Misc`


- Use `Baloon help` if you want to find info on parameters

----

## `Pre-stats`


- Many preprocessing options available - e.g. highpass filtering

----

## High-pass filtering


- Remove low-frequency (slow, < 100s) signal changes
- Important due to _scanner drift_
- Let's look at the effect of HPF

----

1. Open `FSLEyes`

2. Add `fmri.feat/filtered_func_data.nii.gz`

3. View the two time series
###### - View -> Timeseries
###### - Click `Plotting mode` drop-down -> `Demeaned`

4. Find example of high-pass filtering
###### - e.g., voxel coordinate `24,33,15` starts low and drifts upwards in unfiltered image

----

## `Registration`


----

## `Stats`

1. Go to `Full model setup`

----

## Model Set up

### `Explanatory Variables (EVs)`

###### - Each EV models a different effect

----


- View contents of these 3 column .txt files
1. Open previous terminal window
2. Type `ls` and Enter to view files in working directory
3. Type `cat word_generation.txt` and `Enter` (`cat` will print the contents of the txt file)

----

- 1st column = start time of stimulus (s)
- 2nd column = duration (s)
- 3rd column = intensity (1 = on)

----

- Do the same for `word_shadowing.txt`
##### These 3 column txt files contain all the information about the timing of events in the scanner

----

### Convolving


----

### Contrasts

##### Contrasts are the way in which we express questions (alternative hypotheses).

- For example, the question of whether a positive effect exists (that's associated with an EV called β1) can be expressed as β1 > 0.

- Or to test if the effect size associated with one explanatory variable/IV/Predictor is larger than another can be expressed as β1 > β2.

----

### Contrasts & F-tests
- Question: What do you think each 5 contrasts are testing?
- Tip: EV1 = word generation & EV2 = word shadowing

----

#### Contrast Meanings

##### - `Generation`: this tests for when there was greater activation during word generation compared to baseline.
##### - `Shadowing`: this tests for when there was greater activation during word shadowing compared to baseline.
##### - `Mean`: this tests for when the average activation during generation and shadowing was greater than baseline.
##### - `Shad > Gen`: this tests for when there was greater activation during word shadowing compared to word generation.
##### - `Gen > Shad`: this tests for when there was greater activation during word generation compared to word shadowing.

----

### `View design`
###### (If error occurs on class laptops, refer to the above picture)

----


###### - After all that set-up, FEAT is now able to create a model of predicted activity for each explanatory variable (i.e. word generation / shadowing) across time.
###### - GLM will compare this model to actual BOLD activity and reveal those voxels whose activity is well explained by model i.e involved in word generation and/or shadowing.

----

- ##### We fit a model (predicted activity) to each voxel's BOLD respose (actual activity)


----

- ##### This model is built for each task event and it's goodness of fit (how well we predict the actual data) is assessed for each voxel

----

### THIS IS THE GENERAL LINEAR MODEL

----

### `Contrasts & F-tests`

----

## `Post-stats (thresholding)`
- Important for controlling false positives

---

### Running the analysis

#### 2 Options:

- Simply press `Go`
- Or, type `feat <design.fsf>` in to command line.
- ...or view ready-made results!

---

## Viewing the results


----

## Viewing the results

1. View results directory
##### - `Applications` -> `Files` -> `fsl_course_data/ExBox11/fmri.feat`

2. View results report (`report.html`)

3. See the report on registration, pre-stats and stats - these are good sanity checks

----

## Post-stats

###### - `Post-stats` is where the real results are

1. What contrasts show significant efffect?
###### - Each contrast has an associated results image

2. View cluster based statistics
###### - Click on any contrast image

---

### Let's look at the last two results (contrasts Shad>Gen; Gen>Shad) in 3D using FSLeyes

----

### Viewing the results in 3D

1. Open an anatomical image on which to overlay functional results
###### - Add `fmri.feat/example_func.nii.gz`
2. Overlay statistical map of contrast 4 (`fmri.feat/thresh_zstat4; Shad > Gen`) and contrast 5 (`fmri.feat/thresh_zstat5; Gen > Shad`)
3. Apply different colour maps to each
###### - To change a maps colour, select it in the overlay list and then change the colour map at the top

----

4. Load unthresholded versions
- Add `fmri.feat/stats/zstat5`
- not all significant but still potentially useful

---

## Automating this process

- This was a single subject analysis - imagine doing this for 50+ participants
- We can use the command line for more than file navigation - We can use it to automate this process
- Our design file - `design.fsf` - is simply a text file - lot's of potential

---

# Summary
- Hopefully now be acquainted with:
1. Bash commands in Terminal
2. fMRI analysis software FEAT
3. fMRI data/results visualization with FSLEyes
- Code is Key!

----

# fMRI Analysis Resources

----

## Ginette Mumford's FSL FEAT youtube tutorials

https://www.youtube.com/watch?v=lCwewJJPd5U&list=PLB2iAtgpI4YHlH4sno3i3CUjCofI38a-3&ab_channel=mumfordbrainstats

- This is a great series which takes you from start to finish of a task-based fMRI analysis using FSL's FEAT software

----

## Andrew Jahn's FSL videos

https://www.youtube.com/watch?v=9ionYVXUQn8&list=PLOPaMln1VugP5sSKaa_KMmrrVXMJTMrLi&ab_channel=AndrewJahn

- More examples of scripting + FSL

----

## MIT General Linear Model for fMRI

- 3 part lecture series that explains the general linear model nicely. These were the videos that made the GLM and task-design click.
- Theoretical understanding of the principles of the method

----

## FSL's FEAT tutorial

https://www.fmrib.ox.ac.uk/primers/intro_primer/ExBox11/IntroBox11.html

- Example data here

https://fsl.fmrib.ox.ac.uk/fslcourse/lectures/practicals/feat1/index.html

- More explation here

----

https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FEAT/UserGuide

- The master guide to FEAT

----

## GLM Hanbook

https://www.fmrib.ox.ac.uk/primers/appendices/glm.pdf

---

Attendence sheet