error message when trying to reconstruct lead

[mheldmann]

Dear all,
after normalization I am trying to reconstruct my electrode positions. Unfortunately, I get several error messages starting with “Preparing contrasted volume…
Undefined function or variable ‘t’.

Error in nanmean (line 7)
N = sum(~isnan(t), dim);”

Any idea what’s wrong? I am runnning lead-dbs under matlab 2012b, spm 8 and linux Debian.

Thanks in advance, Marcus

[andreashorn]

Dear Marcus,

I’m sorry for this error, most of all because I’ve known about it for some time now and should definitely fix it soon. The problem is that lead tries to use some builtin Matlab functions but instead uses functions supplied with fieldtrip (part of SPM) which have the same name.
The functions in doubt should be located here:
spmfolder/external/fieldtrip/connectivity/

and are called nansum, nanmean and nanvar.

A quick-fix would be to just rename these functions (maybe append a _ft to them or something temporarily) and the whole thing should work again. If you are a fieldtrip user, it might get a bit complicated since I guess that fieldtrip will need these versions for some connectivity analyses.
I will fix this issue in an update soon.

Yours, Andy

[mheldmann]

Thanks for your fast reply. I discovered another minor “typo”. The code refers to “/spm8/toolbox/seg/TPM.nii” when trying to use DARTEL normalization. Unfortunately, in SPM 8 it is written “Seg” instead of “seg”.

Greetings, Marcus

[andreashorn]

Hi Marcus,

thanks a lot for reporting this typo! I have fixed both errors and am uploading an update at the moment.
In theory, there should be an “Update” button in the top row of the lead-GUI and you can press it to download the update.
If not, you can run ea_update(‘force’) to do so.

Did the rest of the process work for you so far?

Yours, Andy

[mheldmann]

Dear Andy, actually I am fighting with the resolution of our structural MRs. All MRs were T1 images recorded with 1 mm isovoxel, I realized from your – by the way very nice – publication, that you recorded T2 with 0.5 mm in 2 dimensions. This makes the detection of the contacts much easier. Greetings, Marcus

[andreashorn]

Okay – I have made the same experience with MPRAGEs before, also T1 doesn’t seem to be as well suited to reconstruct the trajectory, but in general it seems to work for me. In some cases, you can’t delineate all 4 contacts of both electrodes, however, sometimes the uppermost contact seems to show a slightly larger artifact. Since the contacts are equidistant, this might help to place the contacts correctly. Also, you might want to estimate the distance between the electrode contacts, which is usually slightly greater than 2 mm (or 3 mm in 3387 electrodes) depending how much warping has been done in the normalization process. The distance should be displayed in the Manual Review window.
When doing the first reconstructions, I agree that it sometimes feels a bit difficult where to exactly place the contacts, but after about 10 reconstructions, in my experience one gets a confident feeling about where to put them.
If you happen to have some imaging data from 3387 electrodes, it might be easier to start with those as well.
But in any case I’m happy that there were no further technical errors so far.
Best, Andy

[Author]

Posts