Analysis of 2 runs


I have a dataset where reviewers recommended that we analyze the pre- scan and post- scan together. In short, we are using fMRI to see our resting state data.

As a result, the plan is to align each epi to the mean T1 and then proceed with analysis. Further, I have been using slomoco instead of 3dvolreg, and the developers recommend the two not be run together.

My pipeline is as follows:

  1. Run Slomoco
  2. Align epi2anat (mean anat)
  3. Use the following on the slomoc data -subj_id $subj                                           \
        -script proc.$subj -scr_overwrite                             \
        -blocks blur mask scale regress      \
        -copy_anat $anat_dir/anat_ns_mean+orig                          \
        -dsets                                                        \
            $epi_dir_pre/SLOMOCO5/${subj}pre+orig.HEAD  \
            $epi_dir_post/SLOMOCO5/${subj}post+orig.HEAD  \
        -blur_size 5.0                                                \
        -regress_stim_times                                           \
            $top_dir/LeftHand.txt                                 \
        -regress_stim_labels                                          \
            LH                                                       \
        -regress_basis 'BLOCK(45,1)'                                \
        -regress_make_ideal_sum sum_ideal.1D                          \
        -regress_est_blur_epits                                       \

The code runs fine, except I’m left with a "grid mismatch error" of the 2 runs on the .scale data.

My solution was to 3dresample the post- epi to the pre- epi. I no longer received the error, but I’m not sure if this is the right plan of action. I have read on the forum that others have used 3drefit, but from my understanding this doesn’t change the .BRIK file.

When I compare the stats between the two approaches (3dresample vs. not doing anything different) there are slight differences.

Can someone guide me to the right solution?



To start, I have never used the Slomoco program, so I can’t advice on usage/implementation wtih that.

Do your input EPis have different spatial resolution right here?

3dinfo -same_all_grid \
            $epi_dir_pre/SLOMOCO5/${subj}pre+orig.HEAD  \

You could include an “align” block in your as well, to align the EPIs to the anatomical in this command.



The output is 11101. The 0 means they are not on the same grid, right? When I resampled to one of the EPI’s then they are on the same grid (11111).

I checked the original scans that were not aligned to the mean_anat, and they are not on the same grid. For whatever reason i assumed that they would be because the same acquisition was used for the pre and post scan. Does the ‘grid’ refer to the placement of the scan window in the MRI?

Regardless, based on this, I believe resampling was the way to go (or maybe I’m wrong?).