As per Kilner 2005 and advice from the SPM forum I am trying to produce images from individual time-frequency (TF) maps of conditional trial averages in source spaces. Thus for each subject I have conditional TF power values for two virtual sensors (Lt & Rt auditory) for 2400 ms epoch (800ms baseline) data between 5.5 - 84Hz (0.5 freq steps). The paradigm is a simple 2x2 presentation of auditory word (high & low frequency) and non-word (high & low) control stimuli. I used spm_eeg_ft2spm to arrive at the SPM8 friendly input (.mat & .dat). Example SPM structure for a individual is
D =
type: 'evoked'
Nsamples: 2880
Fsample: 1.1996e+03
timeOnset: -0.8000
trials: [1x1 struct]
channels: [2x1 struct]
data: [1x1 struct]
fname: 'r53CtrlHi.mat'
path: ''
sensors: [0x0 struct]
fiducials: [0x0 struct]
artifacts: [0x0 struct]
transform: [1x1 struct]
other: [0x0 struct]
history: [0x0 struct]
D.channels =
2x1 struct array with fields:
label {'MLT'} {'MLR'}
bad
type
X_plot2D
Y_plot2D
units
D.data =
fnamedat: 'r53CtrlHi.dat'
datatype: 'float32-le'
y: [2x158x2880 file_array]
D.trials =
label: 'Undefined'
bad: 0
events: []
onset: 0
repl: 1
I don't neccessarily wish to average across 'channels', nor do I have any strict apriori about frequency bins to average across. When using spm_eeg_convert2images with the current structure format I am forced to choose between averaging across channels or frequency range. Leaving the field blank causes Matlab to hang. Choosing channels to average across causes:
??? Index exceeds matrix dimensions.
Error in ==> spm_eeg_convert2images at 105
images.channels_of_interest = meegchan(selection);
And choosing a frequency range (e.g. [30 50] to average across with default output image dimension (32) and either interpolate/mask option causes:
??? Error using ==> cgprechecks at 47
Not enough unique points specified.
What I hope to achieve is smoothed time-frequency SPM images I can use to carry out FWE corrected stats on. Any help would be much appreciated.
-K.
-------------------------------------------------------------
Kambiz Tavabi PhD
Biomedical Imaging Laboratory
The Children's Hospital of Philadelphia
34th Street and Civic Center Boulevard
Philadelphia, Pa. 19104
Tel: 267.426.0302
email: [log in to unmask]
-------------------------------------------------------------
|