I am trying to extract the sources from labels using
mne_compute_raw_inverse. The command finishes without error, but extracts
extremely few sources for each label: 2 for a V1 label, and 3 for a rather
large fusiform label. Both labels are based on the freesurfer parcellation
of the surface and look fine in tksurfer. The exact command I am dropping
looks like this:
and my output is below. Any help would be appreciated!
Thank you,
Matt
mne_compute_raw_inverse version 1.15 compiled at Aug 19 2011 04:09:29
input file : S011/PM01_clean_raw.fif
inverse operator file : S011/ROI_freq/PM01_raw-5-meg-inv.fif
SNR : 3.000000
Additional channels : 'STI 014'
All label files in S011/ROI_freq/ will be processed.
Source locations will be listed in head coordinates.
Reading the inverse operator...
Read 2 source spaces from S011/ROI_freq/PM01_raw-5-meg-inv.fif with
a total of 11402 source locations
Read the sensor covariance matrix (full)
Read the source covariance matrix (diagonal)
Measurement file id not found (omit matching).
Solution is based on MEG.
Inverse operator information was stored in head coordinates.
Source orientation prior information read.
Depth-weighting prior information read.
Source orientations read.
Singular values read.
Eigenvectors read.
Number of channels = 305
Number of sources = 11402
Free source orientations
Location information in head coordinates
Projection in effect:
# 1 : axial-PCA-01 : 1 vecs : 102 chs MEG active
# 2 : axial-PCA-02 : 1 vecs : 102 chs MEG active
# 3 : axial-PCA-03 : 1 vecs : 102 chs MEG active
# 4 : axial-PCA-04 : 1 vecs : 102 chs MEG active
Adding derived data to the inverse operator (nave = 1)...
Projection applied to the covariance matrix.
Decomposing the sensor noise covariance matrix...
Eigenvalue decomposition had been precomputed.
Eigenleads multiplied with Cholesky decomposition of the source
covariance matrix.
nave change: 1 -> 1
Raw data file S011/PM01_clean_raw.fif:
Processing label directory S011/ROI_freq/...
2 left-hemisphere labels in S011/ROI_freq/
1 right-hemisphere labels in S011/ROI_freq/
Processing left-hemisphere labels...[done]
Processing right-hemisphere labels...[done]
Saving raw estimates...
Data will be split into <= 1907.3 MByte pieces
Processing raw data [2000 samples in each buffer] .........[1000
samples in each buffer] [done]
393000 samples (including skips) saved to
S011/ROI_freq/rh.V1.PM01.normal-mne_raw.fif ... closing...[done]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20110819/de722ee6/attachment.html
Hi Matt,
Did you load the outname_raw.fif in mne_browse_raw to take a look? I
haven't used this command much but I think this command only saves
summary waveforms for each label i.e the average over the labels.. i.e
one time course per label (for fixed orientation inverse) in the
directory.. Not sure why it saved 2 for the V1 label and 3 for the
fusiform label...
If the directory has 2 V1 labels (lh and rh for instance), I'd expect to
get 2 timecourses in the outfile + possibly stim channels..
I also don't have a lot of experience with that command, but looking at your output it seems you have a very decimated cortical surface (only 11000 vertices for the whole brain). This on top of the possible loss of vertices in this region from the distance to BEM calculation could easily leave you with only 3 vertices in the Fusiform and 2 in V1. The easy way to check whether or not there is an error in the program is to load up the forward model in matlab and compare the
forward.src.inuse
to the vertex numbers in the label file (be careful with the starting point of 0 in vertex numbering).
This will let you know whether there is some error going on. If only three of the Fusiform vertices in the label are in the .inuse field than this is only a result of your decimation and possibly the distance to BEM calculation.
Matti and Hari, thanks for explaining the summary waveform option. I removed
the labeldir flag and can extract ~100 sources for each V1 ROI.
Dan, thanks also for your insight. Do you think that a highly decimated
surface will undermine an analysis in any way? What is a more typical number
of vertices for the entire brain?
Matti and Hari, thanks for explaining the summary waveform option. I removed the labeldir flag and can extract ~100 sources for each V1 ROI.
Dan, thanks also for your insight. Do you think that a highly decimated surface will undermine an analysis in any way? What is a more typical number of vertices for the entire brain?
The about 11000 is the total number of sources which is ok. Nothing to worry about there.
- Matti
Thanks again!
Matt
Hi,
With the labeldir option there will be only one summary waveform / label in the output file.
- Matti
Hi all,
I am trying to extract the sources from labels using mne_compute_raw_inverse. The command finishes without error, but extracts extremely few sources for each label: 2 for a V1 label, and 3 for a rather large fusiform label. Both labels are based on the freesurfer parcellation of the surface and look fine in tksurfer. The exact command I am dropping looks like this:
and my output is below. Any help would be appreciated!
Thank you,
Matt
mne_compute_raw_inverse version 1.15 compiled at Aug 19 2011 04:09:29
input file : S011/PM01_clean_raw.fif
inverse operator file : S011/ROI_freq/PM01_raw-5-meg-inv.fif
SNR : 3.000000
Additional channels : 'STI 014'
All label files in S011/ROI_freq/ will be processed.
Source locations will be listed in head coordinates.
Reading the inverse operator...
Read 2 source spaces from S011/ROI_freq/PM01_raw-5-meg-inv.fif with a total of 11402 source locations
Read the sensor covariance matrix (full)
Read the source covariance matrix (diagonal)
Measurement file id not found (omit matching).
Solution is based on MEG.
Inverse operator information was stored in head coordinates.
Source orientation prior information read.
Depth-weighting prior information read.
Source orientations read.
Singular values read.
Eigenvectors read.
Number of channels = 305
Number of sources = 11402
Free source orientations
Location information in head coordinates
Projection in effect:
# 1 : axial-PCA-01 : 1 vecs : 102 chs MEG active
# 2 : axial-PCA-02 : 1 vecs : 102 chs MEG active
# 3 : axial-PCA-03 : 1 vecs : 102 chs MEG active
# 4 : axial-PCA-04 : 1 vecs : 102 chs MEG active
Adding derived data to the inverse operator (nave = 1)...
Projection applied to the covariance matrix.
Decomposing the sensor noise covariance matrix...
Eigenvalue decomposition had been precomputed.
Eigenleads multiplied with Cholesky decomposition of the source covariance matrix.
nave change: 1 -> 1
Raw data file S011/PM01_clean_raw.fif:
Processing label directory S011/ROI_freq/...
2 left-hemisphere labels in S011/ROI_freq/
1 right-hemisphere labels in S011/ROI_freq/
Processing left-hemisphere labels...[done]
Processing right-hemisphere labels...[done]
Saving raw estimates...
Data will be split into <= 1907.3 MByte pieces
Processing raw data [2000 samples in each buffer] .........[1000 samples in each buffer] [done]
393000 samples (including skips) saved to S011/ROI_freq/rh.V1.PM01.normal-mne_raw.fif ... closing...[done]
The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.