PSD and filtering

Discussions
1

Back in July I calculated PSDs for individual trials to extract the values and look at correlations between frequency band power and some behavioural measures. I wanted to do this fresh, so last week I re-ran my bandpass filter on my continuous files (2-30 Hz) and imported the trials, ran the PSD on the sensor data, and extracted it. As far as I can tell, I haven’t changed any projectors for the continuous files or any parameters for the PSD calculation, and this difference exists for all subjects.

However, the power values for the frequency bands I extract are different from those I extracted in July. They’re similar, but on a sensor-by-sensor basis they’re on average higher in my July data. I’m still trying to figure out what might have changed, but I was wondering if I could get a copy of the Brainstorm build from before July 15 2016. I know the bandpass method was recently changed, but I get the same values from a September 23rd build. I just want to rule out a difference in the Brainstorm build before I go back to searching through the continuous files.

TY, Benjamin

2

Indeed, on Sept 23rd, the modifications to the bandpass filters were not pushed yet. The difference must be coming from something else.

About the bandpass filters, if you want to use the old implementation, select the option “Use old filter implementation”.
The new specifications are detailed here: http://neuroimage.usc.edu/brainstorm/Tutorials/ArtifactsFilter#Filter_specifications:_Low-pass.2C_high-pass.2C_band-pass

I’ll send you a link to the version you want by email.

3

Thanks Francois. I’ll also keep looking around to see what else I might have changed since then.