Where the current reaches, and where it does not

What it does

Electrodes are selected on a 3D head model, assigned as anodes or cathodes, and a total current is set; the tool then shows the resulting electric field on the cortical surface. The computation is immediate: no job queue, no waiting.

It is meant for the moment before the experiment, when electrode placement is being decided and two montages need to be compared without preparing a full simulation for each one.

Both montages aim at the same place. The classic one spreads the field over seven times more cortex; the HD 4×1 concentrates it and loses intensity. That is the decision the tool puts on the table before anything is set up.

How it is built

The field of any combination of electrodes is the weighted sum of the individual solutions. That sum is not an approximation: it is exact for the model that was solved.

The foundation is a leadfield: the electric field solution for each electrode on its own, computed once with finite elements and stored. Since the problem is quasi-static and ohmic, recombining is enough.

• Solved with SimNIBS 4.6 on the Ernie head model from its example datasetSimNIBS 4.6
• 10-10 EEG cap (Jurak 2007), with Cz as the reference76 electrodes
• Round electrodes 20 mm in diameter with 4 mm of gel, meshed onto the scalp20 mm
• Field interpolated on the middle gray matter surfacegray matter
• Normalised to V/m per mA, so the chosen current scales linearlyV/m · mA
• Surface reduced from 308,791 to 20,003 vertices so that it fits in the browser20,003 vertices

What it shows

The magnitude is the field strength, regardless of its direction.

The normal component is the projection of the field onto the perpendicular to the cortical surface. It is positive when the field points into the cortex and negative when it points out of it. It is the component most closely related to neuronal polarisation, and therefore the one usually reported.

The three metrics reported with each montage are deliberately conservative:

• Peak: the 99th percentile of the magnitude, not the absolute maximum, which a handful of vertices dominate and which is unstable between meshes99th percentile
• Median: the typical field over the whole cortexmedian
• Focality: the percentage of surface above half the peak. The lower it is, the more focalfocality

Figures are exported as PNG with the caption already written: montage, current, model, reference electrode and electrode geometry. The on-screen value layer can also be turned off.

Limits

• A single head model. The values are indicative, not subject-specific: another skull, another bone thickness or another anatomy give different figuresone model
• The electrode geometry is fixed at computation time. A montage with 5×5 cm sponges spreads the same current over eight times the area and is not reproduced herefixed geometry
• The surface is decimated from about 309,000 to 20,000 vertices: the fine detail of the sulci is smoothed outdecimated mesh
• Cortical surface only. There are no values for deep structurescortex only
• Stored as 16-bit floating point, which introduces a small but non-zero rounding16 bits

How to cite it

SimNIBS, the software that solves the physics behind the tool, should also be cited: Thielscher, A., Antunes, A. and Saturnino, G. B. (2015). Field modeling for transcranial magnetic stimulation: a useful tool to understand the physiological effects of TMS? IEEE EMBC.

The code and the computation scripts are published at github.com/Omega-Collective/omega-efield-lab. To reproduce the leadfield, or if a model with a different electrode geometry is required, write to [email protected].