[FieldTrip] power line noise reduction

[jan-mathijs schoffelen]

Hi Beth,

It seems from your approach, that you use a long stretch of data on which you apply the dftfilter. The dftfilter only works well, if the power line fluctuations are stationary in amplitude. Another way of stating it, is that the line noise really has to be a very narrowband line in the spectrum (relative to the epoch length). Typically, the dftfilter works well only up to an epoch length of ~ 10 seconds (sometimes even less if there are sources of non-stationary power line interference). In your case I would either chop up the data in shorter segments, or use a notch filter (or a lowpassfilter; I would rather go for the low frequency stuff to begin with in the resting state).

Best,

Jan-Mathijs


On Aug 8, 2011, at 6:08 PM, Belluscio, Beth (NIH/NINDS) [E] wrote:

> Hi all-
>  I am evaluating power using ft_freqanalysis for resting state data.  
>  In the preprocessing step, I indicate to remove power line noise at 60 Hz as follows:
> 
>    cfg = [];
> cfg.dataset = 'resting.ds';
> cfg.channel = {'MEG'};
> cfg.continuous = 'yes';
> cfg.detrend = 'yes';
> cfg.dftfilter = 'yes';
> cfg.dftfreq = [60 120];
> dataCB24rest = ft_preprocessing(cfg)
> 
>  Then I evaluate the power at frequencies from 5 to 100 Hz as follows: 
> 
> cfg = [];
> cfg.method = 'mtmfft';
> cfg.taper = 'hanning';
> cfg.foi = 5:5:100;
> cfg.tapsmofrq = 4;
> cfg.output = 'pow';
> freqCB24rest = ft_freqanalysis(cfg, dataCB24rest)
> 
> When I visualize the result, I get a VERY large value for the power at 60 Hz in every channel.  For example, see attached.
> 
> What am I doing wrong?
> Thanks for your help,
> Beth.
> 
> 
> Beth Belluscio MD-PhD
> Clinical Fellow
> Human Motor Control Section
> NINDS, NIH
> 301-402-3495
> 
> 
> <example.tif>_______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip

Dr. J.M. (Jan-Mathijs) Schoffelen 
Donders Institute for Brain, Cognition and Behaviour, 
Centre for Cognitive Neuroimaging,
Radboud University Nijmegen, The Netherlands
J.Schoffelen at donders.ru.nl
Telephone: +31-24-3614793