DUNEuro error with the tutorial dataset

tmedani · February 15, 2022, 1:55am

@Francois did we fixed this error with the last discussion with iso2mesh dev team?

FEM tutorial: Issues to fix

opened 09:57AM - 13 Aug 21 UTC

tutorials

1) **SimNIBS**: Remeshing the SimNIBS model with iso2mesh doesn't work. Here is …the error message: ``` FEM> 1. scalp: C:\Work\Protocols\TutorialFem\anat\Subject01\tess_scalp.mat FEM> 2. skull: C:\Work\Protocols\TutorialFem\anat\Subject01\tess_skull.mat FEM> 3. white: C:\Work\Protocols\TutorialFem\anat\Subject01\tess_white.mat FEM> 4. gray: C:\Work\Protocols\TutorialFem\anat\Subject01\tess_gray.mat FEM> 5. csf: C:\Work\Protocols\TutorialFem\anat\Subject01\tess_csf.mat generating tetrahedral mesh from closed surfaces ... creating volumetric mesh from a surface mesh ... *************************************************************************** ** Error: Line 117: surf2mesh (line 117) ** Tetgen command failed ** ** Call stack: ** >surf2mesh.m at 117 ** >process_fem_mesh.m>Compute at 442 ``` 2) **Pre-processing**: The PSD screen capture does not seem to be from this file (not the same time definition) 3) **Averaging**: The averaging screen capture shows a weighted average: this is confusing => Uncheck the "weighted" box 4) **Noise covariance**: - The name of the epochs in the screen capture is wrong ("2" and not "STI101") - The is a mention to a DC offset removal that was not applied when importing the epochs - The time window doesn't match between the text and the screen capture. 5) **Inverse model**: Redo screen captures with correct average file name. 6) **Dipole scanning**: - Redo screen capture to show only the LINKS (we can't select the shared kernels in Process1, as the text said before I edited it) - The dipole scanning process is executed over 40-60ms, but the results are illustrated at 21ms... - On my end, peak is at 22ms, not 21ms - If you scan the dipoles at multiple time points, you need to explain in the tutorial how to navigate in time and/or overlay multiple time points. Otherwise, simply scan at one specific time sample (eg. 21ms), this is simpler to display and post-process. - I find the second figure confusing. It should be split in two: in this section, show only the merge+renaming of the dipoles. Then make another figure to illustrate the dipoles display, and move it to the following section (P20/N20) 7) **Comparison: DTI vs isotropic**: - Add text to explain that you change the coloring of the dipoles "by group" - Explain how to change center the slices of the MRI 3D on a dipole (click on the dipole, the click on the "View/3D" button in the toolbar of the Dipole info panel) - "When we compare between the EEG and the MEG, there is a difference of 20mm between the localization between the MEG and the EEG": It looks less than 2cm in your screen captures... It would be worth illustrating how to compute this correctly: click on each dipole to see their locations (add screen capture of the dipole info panel), and compute the norm of the difference of the two vectors (`sqrt((x1-x2)^2+(y1-y2)^2+(z1-z2)^2)`) from the Matlab command line to get the exact distance (screen capture of the Matlab command window). - You need to explain the granularity of these results: you are using the cortex as the source space, the output of the dipole scanning can only be one of the vertices of the cortex surface. The two methods lead to the selection a selection of difference vertices, but it doesn't mean that there is "an error of XX millimeters" between the two methods. - "as well as a difference in the orientation in the coronal view": No, this difference in orientation is clearly visible in the three views. You can also compute the angle between the two from the orientation vector read in the dipoles info panel. 8) **Comparison: MEG vs EEG**: - "If we compare the difference between the EEG and MEG dipoles": Do you merge again the dipoles in a different way for this? If so, you need to tell the user to repeat the operation (possibly, this should be added in the previous section "Dipole scanning", as this is where you did the other merge). => Or you can keep the section but skip the figure , as it doesn't bring any additional interesting information (left and right figures look too similar) - I'm not sure what you can see here that you can't see in the in the previous figure. 9) **Comparison: FEM vs BEM**: - I get different plots than yours for EEG FEM DTI vs. EEG BEM: ![image](https://user-images.githubusercontent.com/6920058/129355933-e4f7c603-1a93-47e2-8b9c-0369cc52bf19.png) ![image](https://user-images.githubusercontent.com/6920058/129355979-02d4fd32-29b5-4e59-b0ce-992aa14b7403.png) ![image](https://user-images.githubusercontent.com/6920058/129356033-1a74859a-ddef-49fc-b2af-8630b04a5952.png) 10) **DUNEuro advanced options**: Merged into the DUNEuro tutorial: https://neuroimage.usc.edu/brainstorm/Tutorials/Duneuro#DUNEuro_options:_Advanced_.5BTODO.5D => Must be edited, most options are not documented. 11) Removed text (too drafty): "Dipoles outside of the grid for the MEG: remove this dipole or replace it with its spatial neighbors (not the neighbor in the list) ==> not easy to handle write small tuto? " 12) Review associated FEM tutorials (Francois): - https://neuroimage.usc.edu/brainstorm/Tutorials/Duneuro => **DONE** - Documentation of all the advanced options missing: - https://neuroimage.usc.edu/brainstorm/Tutorials/FemTensors => **DONE** - https://neuroimage.usc.edu/brainstorm/Tutorials/FemMesh => **DONE** - Iso2mesh: Removed link "From our test", which was pointing at https://neuroimage.usc.edu/brainstorm/reviewMEEGForward?highlight=(duneuro), which does not exist. - Juan, Takfarinas: Please have a look to make sure that it still looks like what you wanted.

I believe that the error is due to some hole on the mesh, but I will check it again.