Dear list,
I obtained some responses by convolving box-car like paradigm (as
descrbed in IEEE Trans. Med. Imag, 19:1188) with different filters
(poisson/gaussian/gamma/canonical HRF) plus some random noise, and
hope to test SPM's performance on them. To let SPM understand the
simulated data, I generated 32x32x32 array, and set the obtained
responses to two voxels and random noise to other voxels. I used the
f.write.analyze function in AnalyzeFMRI package in R
(www.r-project.org) to write the simulated data into a series of .img
files (sim.0001.img, sim.0002.img ...) - each .img file representing a
3D plot in one time point.
Though there is no problem in specifying the model, however, the model
estimation seems no working. (The sample auditary dataset works
perfectly). A warning poped up and said "no inmask voxels - empty
analysis". I googled this message and only found several pages
linking to SPM source codes. Then I checked the original source code
(cuixuprocess.m) but cannot figure out the problem.
I wonder why it was failed to let SPM analyze the simulated data in
.img files.
Thanks,
Wuming Gong
|