Chapter #5: Preprocessing
Preprocessing a resting-state dataset is similar to how you preprocess a task-related dataset: You do the same steps of realignment, segmentation, normalization, and so on. The main difference is in the motion threshold. Resting-state datasets are notoriously susceptible to motion-related artifacts; even small movements can introduce spurious correlations. These artifacts are particularly problematic for laboratories studying disorders such as schizophrenia, or for researchers who scan very old or very young people. Too much movement will ruin your study.
There is no simple way to address this problem, aside from training the subjects to move as little as possible. That said, the CONN toolbox has several methods for mitigating movements artifacts, including ART (Artifact Detection Tools) and principal component filtering of signal from tissues you may not be interested in, such as white matter and cerebrospinal fluid - a method known as
Comp Cor, short for Component Correction. In the following chapters, we will examine how including or omitting these steps will affect our results.
Click on the
Preprocessingbutton again, and note that all of the previous preprocessing steps have already been loaded. Select
Add, and note all of the options that are presented for processing both the functional and structural data. Find the preprocessing step to remove the initial scans, and then click the
Move upbutton to move this step to the top of the preprocessing list. Click
Start. When prompted for the number of scans to remove, enter
2. Keep the rest of the settings as the defaults, except for Smoothing, which you can change to 4mm. Click
OK. When the preprocessing for this subject finishes, click on
Functional, and take a screenshot of the resulting image. Similar to smoothing functional images for task-based data, the volumes should look different. When would you want to use a smaller smoothing kernel as opposed to a larger smoothing kernel?
For a video overview of how to preprocess data in the CONN toolbox, click here.
If everything has run without error, you should see a pop-up window saying that everything has finished without any problems. When the preprocessing has finished, we will need to inspect the images for any artifacts or other problems - in other words, we will do Quality Assurance (QA) checks. To learn more about how to do them, click the