Dear Qiwen,

>
> I have a couple of questions for you, I would appreciate it very much if you would give any comments, suggestions, or information.
> The following is the method that I used to get statistical maps about one of my block design fMRI study. I used the realignment parameters file (realignment_motion parameters_.*txt ) as a set of 6 covariables to regress out movement-related activations.
>
> The study is a Block Design, TR=2.27583 seconds, the condition for successive blocks alternated between MAS and REST, starting with MAS, the whole functional EPI scan comprises of 15 cycles, each cycle contains 2 epochs, MAS and REST. MAS takes 6 scans(timepoints) and REST takes 18 scans(timepoints). The total scans (timepoints) is 15*24=360.
> After realignment, slice timing, spatial normalization, and smoothing the data, the statistical maps were generated in the following way,
>
> Press ?fMRI model?
> Pull down menu: Would you like to? ?specify a model and estimate the parameters?
>
> Interscan interval: 2.27583
> Scan per session: 360
> Conditions or trials: 1
> Name of trials 1: MAS
> Stochastic Design: no
> SOA: Fixed
> SOA for MAS: 24
> First trial(scans): 0
> Parameter modulation: none
> Are these trials : epochs
> Select type of response: fix response (Box-Car)
> Convolve with HRF: yes
> Add temporal derivatives: no
> Epoch length (scans) for MAS: 6
> User specific regresors: 6, (spm_load the parameters txt. file generated during realign process).
> Remove Global effects: Scale
> High pass filter? Specify
> Cutoff Period:109( =2*24*2.27583)
> Low-pass filter: hrf
> Model intrinsic correlations: none
>
> Results:
> Select ?t-contrast?
> Type a ?1 0 0 0 0 0 0? in the contrast area to detect positive activation
>
> Type a ?-1 0 0 0 0 0 0? in the contrast area to detect negative activation
>
> Question 1: Some people say the contrasts should be ?1 1 1 1 1 1 1? instead of ?1 0 0 0 0 0 0? to get the positive map which will regress out the motion-related activation (three translations and three rotations), How do you think of this?
>

I wouldn't hang out with those people any more. Use the [1 0 0 0 0 0 0] (which by the way you can enter as [1]) contrast weight vector.

>
> Question 2: Does the realignment_motion parameters_.*txt tell me how much motion movement was detected? or was removed?

Good question. It tells you how much movement was detected, and it does so with hich accuracy. The process of finding the movement is called "realignment". The second step: recreating the images as if they had been scanned in a different position is known as "resampling" or "reslicing". Unfourtunately this is a bit trickier, especially for fMRI, and the resulting (resliced) time-series will often still contain considerable movement related variance.

So in principle

1.SPM detects the movement (realignment)
2. SPM attempts to remove the movement (reslicing)
3. Careful neuroimagers include the movement parameters in the design matrix to remove any residual movement related variance.

>
>
> Question 3: from spm99, after getting the group statement, how can time-course of each voxel (intensities vs. time-points) be extracted based upon specific cluster so that the signal percent change of each voxel can be calculated and the average time-course of each cluster across subjects can be also generated? (I 've got the group statement(final mask). Given specific P value, such as 0.001, the corresponding cluster-mask was generated. My question is that how can I use the Cluster-mask back to each subject's functional scan, and calculate the intensities of each voxel vs. time point in each cluster in each individual scan and produce a .txt file then I can put them(intensity of each time point of each voxel in a specific cluster) to Excell to calculate the signal percent change. I know, it's easy to do that using the command "3dmaskdump" from AFNI, I do not know how to do that using SPM 99.)
>

There is no direct way to do that in SPM. It is easy enough though with a bit of Matlab programming. Also, I think Christian Buchel might have something like this as part of his SEM toolbox.

>
> Question 4: Do you know how to convert img and hdr files of SPM99 to BRIK and HEAD files of AFNI?
>

Sorry, no.

Good luck Jesper