Geometric void (diamond-shaped hole) in PSD Source display (sLORETA)

Discussions
1

Hi Brainstorm Team,

I am working on an EEG study (64 channels, BrainVision format) regarding the effects of cTBS. I am encountering a persistent visualization issue when displaying the Power Spectral Density (PSD) on the Cortex surface.
Data & Anatomy:

  • Anatomy: Individual MRI processed with CAT12. 3-shell BEM surfaces generated (no intersections).

  • Electrodes: 64 channels with digitized positions (.xyz file), correctly projected on the scalp.

  • Processing: 120 epochs (2s each) of resting-state EEG.

The Issue: When displaying the Welch PSD (Average across trials) on the Cortex (specifically in the Alpha band, a geometric, diamond-shaped void appears in the center-parietal region (near Cz/Pz). This "hole" has no data or is rendered black/transparent, even though the electrodes in that area have good quality signal and are correctly labeled.

Steps already taken:

  1. Verified channel names (matching header vs. .xyz file).

  2. Tried both OpenMEEG BEM and 3-shell sphere forward models.

  3. Re-projected electrodes to the scalp surface.

  4. Computed sLORETA sources first, then Welch PSD on sources (Average across trials).

  5. Increased "Max distance" for interpolation in the Surface tab.

Any advice on how to "fill" this gap to ensure valid ROI extraction for my dissertation would be greatly appreciated.

Thank you for your amazing work on this toolbox!

image
image750×667 50.2 KB

2

Adding more images because I couldn’t since I’m a new user.

image
image810×620 63.1 KB

3

Hi @Madalina

Do you have the same issue when displaying the EEG (EEG top plot with 2d or 3d)?

4

Hi @Madalina,

There are some points that are not clear in your message.

It is mentioned that the issue is:

However, in the figure seems to be the PSD on the sensor data, that is why the electrodes are on that surface (the EEG cap), otherwise it would be wrong, as the EEG electrodes are not on the cortex.

The holes in the representation suggest that there are some NaN values.
How did you compute the PSD?