 |
 |
Dear Nadine, I'll try to answer this ... > Hello list! One of our colleagues is planning a motor sequence task > (sequence will be displayed visually and subject has to reproduce > sequence) between a group of patients and a group of healthy > controls. For our experimental hypothesis it will be crucial to > have patients perform the motor task at their own speed and we > would measure speed outside of the scanner beforehand and then > adjust either the block duration or number of stimuli according to > this speed. > Here are my questions: > > a) Would it be better to have the two groups perform the equal > amount of stimuli (e.g.; each group does 5 motor sequence > reproductions per block but the block length will differ between > groups) or would it be better to have the same block length and the > number of stimuli will be greater in the healthy controls > (therefore more manually produced sequences and more visual stimuli > etc. in the controls)? > > I am just not sure what will introduce a bigger group bias > ...varying the block duration between groups or varying the number > of performed motor sequences in one block. Or is this kind of > design just impossible? I would suggest keeping the block length the same (see below why) and have a different number of stimuli. When you model the blocks, it doesn not matter how many events you have, only the block duration (but it does if you model each event to create a block - there is a ppt from Cathy Price somewhere on the SPM website on this). An additional thing you can do is to add a parametric regressor so that for each block you enter the number of responses and therefore the block regressor will fit voxels showing a constant level of activation (say e.g. SMA) vs. the parametric regressor which will fit voxels where blocks amplitude change with the number of response (a priori motor cortex). The cool thing here is that if you have a difference between groups for the block regressor, you can be quite confident that it is not related to the number of events (since it is regressed out). > b) Is it even possible to compare two groups of patients with > different block durations (e.g.; 18 seconds versus 32 seconds) or > would that lead to increased activation within the group with a > longer block sample? Has anyone tried this? Any references I could > look up? I would not use different block durations simply because the power of the design would be different between groups (in SPM --> review design --> explore: you can see the regressors in the frequency domain) - and therefore I would think that results will be biased. > c) Maybe an event-related design would be better here but > then the trial duration would change between groups? Has anyone > tried this? Maybe a self-paced event-related design? self paced could do the trick and you can get the same number of event per subject :-) Hope this helps Cyril -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.