 |
 |
| The result from our fMRI experiment was far from expected especially for | the second step. T1 and the mean of realigned EPI images can't map exactly | together. Instead, only part of the mean of realigned images (8 slices for | one image file, 62 images) is mapped into T1 image. I doubt if it will | affect the following results. Could someone help me solve the problem in | order to continue the pre-processing? When you say that the mean EPI and T1 image cant map together exactly, do you mean that the field of view of the images were such that there was very little overlap between them, or do you mean that the algorithm failed or that the distortions in the EPI images could not be adequately modelled by a six parameter rigid body transformation? There is not a lot that can be done if the FOV of the images means that there is not much overlap - simply because the data is not there. If the coregistration has failed (use <Check Reg> to find out), then you may wish to try the mutual information registration approach (you can get this by the <Defaults> button). An alternative approach to spatially normalising EPI images would be to match them to the EPI template - or even create your own template. To do this, you would estimate the spatial normalisation parameters from the mean EPI image, and apply the transformation to the individual images (andpossibly also the coregistered T1 image). All the best, -John %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%