 |
 |
> John - I changed the T1 origin to the AC point before any > transformation/registraion was estimated. I did it to try and ensure > optimal segmentation as suggested in previous posts. I did not change > origin in any images follwoing the 1st step - Segmentation. If my > understanding is correct the ROI will be transformed from MNI tempate to > the native T1 (using seg_inv_sn.mat > ), taking into account differences in the origin, and so all should remain > in alignment from that moment onwards. I guess I was worried that a > problem might occur if the origin of the DWI image was not on the AC point > as is the origin of the T1. I could test this by coregistering and > reslicing the T1 and the ROI into DWI space and then applying the same > origin change to the T1 and the DWI image, followed by segment and warp and > write. Sorry, I should have taken more notice of your question. > > Just to recap, the process i followed was: > 1. change origin of T1 to AC point > 2.segment T1 > 3. apply inverse sn paremeters to ROI in MNI space bringin it into native > T1 space > 4.coreg and reslice T1 and ROIs to DWI image I really can't figure out which step went wrong. I would suggest checking which stage the alignment was lost by e.g. 1) Check that the original ROI is in MNI space (Check Reg). 2) Try inverse normalising an MNI brain to see if this matches the T1. 3) Check that the inverse normalised ROI was included as "other" in the coregistration. Maybe you could repeat it all again and see what happens if you check various alignments after each step. Best regards, -John