Every row in the lead field matrix (aka gain), located in the Gain field in the HeadModel structure, has the proper units for the channels: [T] for magnetometers, [T/m] for gradiometers and [V] for electrodes, as it was mentioned: Compute MEG signal with Lead Field matrix - #2 by Sylvain
Yes, those wo sensors have different units. Signals in the gradiometers are converted from [T/m] top [T] ONLY for visualization (for example when plotting time series). Check these links:
No, there is no need of scaling. The lead field matrix has already the proper units.
You can prove this by yourself:
-
Import in Brainstorm MEG recordings with magnetometers and gradiometers
-
Export those recordings to Matlab and plot them with
plot(). You will notice that multiplying the gradiometers signals by 0.04 will give you the same plot as in Brainstorm. -
Compute a head model and compute sources for that recording.
-
In Matlab, simulated MEG (mag and grad) from the sources by multiplying the lead filed by the sources.
-
Plot the simulated signals as in Step 2. Note that the signals are in their proper units.
The function process_simulate_recordings.m contains the code to simulate sensor recordings from sources, you can use it as reference: brainstorm3/toolbox/process/functions/process_simulate_recordings.m at master · brainstorm-tools/brainstorm3 · GitHub