Hello

I am trying to register two EPI datasets. The first dataset is a localizer with 20 slices and the second is the actual experiment with 10 slices which cover only part of the slices of the localizer dataset. The partial coverage is in the axial plane and the x and y dimensions of both datasets are the same. The voxel sizes of both datasets are also the same.

At the time of acquisition, the middle slices of both datasets are shifted 4mm in the z-direction.

In order to register these datasets I tried two options and neither of them provided a good registration.

Option 1: Register the EPI dataset to the mean of the Localizer with 3dAllineate

3dTstat -mean -prefix LocalizerMn Localizer+orig
3dAutomask -prefix EPI_mask -clfrac 0.4 EPI+orig
3dAllineate -base LocalizerMn+orig -1Dmatrix_save EPI2LocalizerMn_Matrix.aff12.1D -prefix EPIReg -warp shift_rotate -warpfreeze -replacebase -lpa -source EPI+orig -source_mask EPI_mask+orig

However this option creates motion artifact across the EPI subbricks as if too good registration was forced for each subbrick with respect to the mean of the localizer.

Option 2: Register the EPI subbricks to their mean with 3dvolreg, register this EPI mean to the Localizer Mean with 3dAllineate, concatenate the matrices of both registrations, and apply this transformation to the EPI subbricks
3dTstat -mean -prefix EPIMn EPI+orig
3dvolreg -Fourier -verbose -base EPIMn+orig -prefix EPIReg -1Dfile EPIRegParam.1D -1Dmatrix_save EPIRegMatrix.aff12.1D EPI+orig
3dAllineate -base EPIMn+orig -1Dmatrix_save EPIMn2LocalizerMnMatrix.aff12.1D -warp shift_rotate -lpa -source EPIMn+orig -source_mask EPI_mask+orig
cat_matvec EPIMn2LocalizerMnMatrix.aff12.1D EPIRegMatrix.aff12.1D -I > temp_Matrix.aff12.1D
3dAllineate -1Dmatrix_apply temp_Matrix.aff12.1D -master LocalizerMn+orig -prefix EPIReg -source EPI+orig

However this option does not work either and having looked at the results a rough shift in the z-direction of 4mm is still required.

I also tried to first shift the EPI datasets with 3drotate for the difference in origin of 4mm,
3drotate -ashift 0 0 4S -rotate 0 0 0 -prefix EPI_shifted.hdr EPI.hdr
but one of the slices of the EPI disappears with just scattered points.

I am sure this must be simpler than I think and any help will be much appreciated.

Thanks very much in advance.
Cesar