Dear experts,
I am analyzing a task-based, event-related design in a sample of 44 individuals with multiple sclerosis. I’m hoping you might have some feedback on the best way to include lesion load maps in my dataset. In previous publications, I’ve included lesion maps as a voxelwise regressor in my final higher-level model (4D file with each image corresponding to each participant).
However, I now see that I can model my HRF at the first-level by including lesion maps. I have several questions to this end that I was hoping to have your input on.
1)Do you believe this would be a theoretically more sound way of modeling out lesion-related activation patterns?
2)Would this save us degrees of freedom in the higher level contrasts?
3)In my first effort to construct a FEAT template using an individual lesion mask as a voxelwise EV, I received an error saying my lesion contrast was out of bounds on the time index. I assume this is because my input was a 3D file, and not 4D. Would multiplying my original lesion mask by the number of volumes in my time series, and merging those images into a 4D file resolve this issue while conserving the integrity of the analysis?
4)Right now I have my lesion masks in functional space for input to the first level voxelwise analysis. Is this correct? Or do they need to be in standard space?
5)How should the lesion masks be weighted? Right now I’m following the format used in setup_masks (lesions valued at 1, all other voxels are 0).
Thank you very much for your time, and have a good evening.
Best,
Alisha
|