 |
 |
Hi Chris, >Long winded version: We often use a fiducial marker to ensure we never >left-right flip an image. I recently used a marker that shows up in the EPI >T2* scans as well as in the T1 anatomical scans. Image preprocessing appears >to proceed smoothly - the realignment and normalization look very sharp. > >However, I notice that the EPI unwarping image that SPM2 generates clearly >includes the marker in its map. This sounds potentially bad, as the marker >will not move when the head moves. > > > Long winded answer: Does that mean the marker is attached to some headholder rather than to the head? If so, yeah I think it will be a (slightly) bad idea when you encounter subjects that move a bit more. I suspect both realignment and Unwarp will get its "knickers in a twist" (just learned that expression, and just had to use it). When estimating movements realign will attempt to minimise the sum-of-squared differences between some reference scan and another scan. In addition, realign tries to be extra clever by preselecting a subset of voxels where movement will have the largest impact (and hence also contains the most information pertaining to movement). In short, the larger the image gradient and the further away from the center of the volume, the larger impact a voxel will have on the estimation. Your fiducials will therefore have a quite large impact, and tend to tell realign "Hey, look at me. There were no movement". Hence there is a risk that movements will be underestimated. As for Unwarp, it will surely get confused (locally). If the marker is far away from the brain, and you have enough basis-functions, the effects may not extend into the actual brain. I wouldn't bank on it though. Good luck Jesper