Hi! I’m a beginner in fMRI data analysis, and I’ve just started learning about data preprocessing and univariate analysis. I used fMRIPrep to preprocess my data and plan to use SPM for first-level and second-level analyses.
I’d like to ask:
Can the data obtained from fMRIPrep be directly used for analysis, or are additional steps like smoothing required?
How can I evaluate data quality, e.g., what thresholds should I use?
Lastly, I’d appreciate any recommendations for learning resources on workflows for different types of analyses.
You usually need to smooth data, at least if you are doing traditional mass-univariate voxel-wise tests. Though pretty much every analysis software has that functionality built in, and shouldn’t be something you have to do separately beforehand.
This depends on many factors, including the amount of data you collect per subject, number of subjects, and population. There is no standard.
No standard seems to be the most uncertain and difficult part to me My current data is like this, there are two groups of subjects, about 25 people in each group, and they have been scanned twice, pre-training and post-training. May I ask what kind of range of dataset do you think is reasonable based on your experience? Thank you!
I don’t know enough about your experiment to give an exact recommendation. You might consider doing a power analysis to see how many subjects you would need to resolve effect sizes that you expect in your study (based on similar literature).