CTF 3rd order gradiometers correction

Robert Oostenveld r.oostenveld at FCDONDERS.RU.NL
Fri Jul 11 10:11:10 CEST 2008

Hi Michael,

On 10 Jul 2008, at 14:05, Michael Wibral wrote:
> (1) Go for the raw data and increase the artefact rejection
> thresholds considerably (as beamforming should not localize
> artefacts inside the head anyway).

Beamforming tries to optimally supress the noise. That means that
noise will not be perfectly suppressed. If you can use the 3rd order
gradient to already reduce the impact that environmental noise has on
your data, you'll make it easier for the beamformer to suppress the
remaining environmental and physiological noise.

With 3rd order gradients switched on, it probably will also be easier
to identify physiological artifacts (eye blinks, muscle) and to
discard those pieces of data from your analysis. Again, if you can
remove noise from the data, the source reconstruction will perform

> (2) Wait for a fieldtrip version where leadfield can be computed
> for 3rd grads (if that's planned at all...).

That is what is now possible, although it requires
cfg.headerformat='ctf_new', cfg.dataformat='ctf_new' and the ctf p-
files (from ftp://ftp.fcdonders.nl/pub/fieldtrip/external). Through
the CTF p-files the balancing coefficients are read and stored along
with the data. The denoise_synthetic function applies the balancing
coefficients to the data _and_ updates the sensor array description
so that the balanging will be taken into account during the forward

If you want to see how it works: do preprocessing and have a look at
data.grad.balance, which contains the details. Then do
denoise_snthetic and compare data.grad.tra prior to and after
applying the 3rd order synthetic gradient. The grad.tra matrix
describes how all individual coils of the MEG system are combined
into channels, and in the comparison you can see that for 3rd order
data the reference coils are also used, whereas for raw data the
reference coils are weighted with zero.

> (3) try to use laedfields from CTF for these data, if that's possible.

that is not needed, see above.

best regards,

PS I suggest to make an example matlab page at http://www2.ru.nl/
id=fieldtrip:documentation:examples:denoise_synthetic. Feel free to
add code and explanation there for other people.

PPS I am curious to hear from other people on how using 3rd order
gradient influences their results

The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/fcdonders/fieldtrip.

More information about the fieldtrip mailing list