Dear Gabriella,
what are we looking at here? Are these con or beta images, residual images or something else? Are the images spatially normalised, smoothed etc? If the images are normalised, what method and parameters did you use? What are the actual voxel values? First you should trace back the sources of striping to your raw, unprocessed data and the acquisition process. A simple, but very effective measure would be to compute mean and standard deviation images for your raw time series using imcalc. In the imcalc batch module, select the original timeseries and set the following items:
- Expression: mean(X) for mean image, std(X) for standard deviation
- Options->Data Matrix: Yes
- Options->Interpolation: Nearest neighbour
Then, look at the resulting images and assess the size and extent of the effect. You may want to repeat this for intermediate time series data (realigned, slice timing corrected, normalised).
If striping is already present in your original data and you have a chance to modify your image acquisition protocol, you should do so. If it appears after a certain processing step, you should review the parameters of this processing step (e.g. interpolation options, smoothing).
Best regards,
Volkmar
|