Exporting GSR signals for analysis

Post Reply
pcp15srl
Posts: 9
Joined: Wed Aug 09, 2017 11:39 am

Exporting GSR signals for analysis

Post by pcp15srl »

Hi,

I realise that this has been covered previously, but I am having trouble understanding what calculations are necessary to convert my GSR signals for analysis in other software. I collected my data using the 16 Hz GSR module as part of my ActiveTwo hardware.

The software I will be using for analysis requires the GSR signals to be in microsiemens. I have exported the GSR.bdf files to text files (containing the signal values in nanosiemens), and have applied the following calculation to the signal: (GSR_value*31.25) / 1000.

Am I correct in thinking this is how I can convert the GSR signals from Biosemi to microsiemens? If not, could someone let me know what the correct calculation would be?

Thanks,

Sam

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Exporting GSR signals for analysis

Post by Coen »

The above formula is correct. As an extra check, compare the values calculated from the BDF file with the numerical values indicated in ActiView during recording (top of left sidebar).

Best regards, Coen (BioSemi)

pcp15srl
Posts: 9
Joined: Wed Aug 09, 2017 11:39 am

Re: Exporting GSR signals for analysis

Post by pcp15srl »

Hi Coen,

Thanks for your response. When I apply this calculation I am getting values that are still quite large (see example of one participants data below), this general pattern of values in the hundreds appears to be consistent across all of my participants, which makes me think there is something wrong with the way I am converting the data. I've plotted the shape of the data across time and it looks sensible, it just appears that the values are off somehow.

Do the values recorded in the original .bdf file correspond to skin conductance level, or some other measure?

Time (s)
0
0.04
0.08
0.12
0.16
0.2
0.24
0.28
0.32
0.36
0.4
0.44

GSR1
911.6961656
911.8223281
912.2758781
912.4184563
912.8480844
912.8633438
912.8616344
913.3444219
913.8662719
914.2657469
914.1261
914.1753531



If you have any more ideas about this I'd appreciate it!

Thanks,

Sam

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Exporting GSR signals for analysis

Post by Coen »

GSR has an LSB value of 0.03125 nanoSiemens.

To convert to microSiemens, devide the value in the BDF file by 32,000 (sorry for introducing confusion with my mistake above).

For example:

Z=100 kOhm, ActiView displays 10,000 nanoSiemens, the BDF file has a value of 320,000 (= 10 uS)
Z=10 kOhm, ActiView displays 100,000 nanoSiemens, the BDF file has a value of 3,200,000 (= 100 uS)
limit: Z=3.815 kOhm, ActiView displays 262144 nanoSiemens, the BDF file has a value of 8,388,607 (2^23-1)

Best regards, Coen (BioSemi)

pcp15srl
Posts: 9
Joined: Wed Aug 09, 2017 11:39 am

Re: Exporting GSR signals for analysis

Post by pcp15srl »

So the only conversion of the values should be GSR_Value/32000?

Here is the same data with this conversion applied:

Time (s)
0
0.04
0.08
0.12
0.16
0.2
0.24
0.28
0.32
0.36
0.4
0.44

0.970605469
0.970840638
0.971307372
0.971433716
0.971761781
0.972013488
0.972329528
0.972667969
0.972893678
0.973229859
0.973488403
0.973714538

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Exporting GSR signals for analysis

Post by Coen »

1 microSiemens conductance is 1 MOhm impedance (two electrodes in series). 500 kOhm per electrode sounds about right on the dry skin at 16 Hz.

Using electrode gel (but no skin scrubbing) should lower the resistance to something like 100 kOhm (10 microSiemens).

Best regards, Coen (BioSemi)

pcp15srl
Posts: 9
Joined: Wed Aug 09, 2017 11:39 am

Re: Exporting GSR signals for analysis

Post by pcp15srl »

Hi Coen,

Thanks for this, I've had a look at the information stored in the .bdf header, so what you've said makes much more sense to me!

Best wishes,

Sam

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Exporting GSR signals for analysis

Post by Coen »

Just a further remark about the range or the GSR measurement:

As said above, the maximum conductance that can be measured is approx. 262 microSiemens (3.8 kOhm)

ActiView has a lower conductive limit of approx. 0.6 microSiemens (1.7 MOhm). At conductances lower than 0.6 microSiemens the green "GSR in range" lights in ActiView go off and a zero value for the conductance is displayed and also stored in the BDF file. So, the BDF file cannot contain conductance values between 574 and 0 nanoSiemens.

Best regards, Coen (BioSemi)

pcp15srl
Posts: 9
Joined: Wed Aug 09, 2017 11:39 am

Re: Exporting GSR signals for analysis

Post by pcp15srl »

Hi Coen,

Yes so this is what is making me think that the conversion is not quite correct. For one of my data files applying this conversion will give you the following values:

0.327120788
0.327234528
0.327570769
0.327817259
0.327959778
0.328092834
0.328210144
0.328242156
0.328274169
0.328317566
0.328340056

Presumably if dividing the .bdf file values by 32,000 converts the values to microsiemens then these results would be zero values?

Best wishes,

Sam

Coen
Site Admin
Posts: 1124
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands
Contact:

Re: Exporting GSR signals for analysis

Post by Coen »

Test with 2 buckets filled with salt water. Put CMS, DRL and one GSR electrode in one bucket, put the other GSR electrode in the other bucket. Connect the two buckets with a 100 kOhm resistor (solder pieces of wire to the resistor, put the bare wire ends in the water). ActiView will show a conductance value of approx. 10,000 nanoSiemens. The BDF file will show a value of approx. 320,000.

Best regards, Coen (BioSemi)

pcp15srl
Posts: 9
Joined: Wed Aug 09, 2017 11:39 am

Re: Exporting GSR signals for analysis

Post by pcp15srl »

Hi Coen,

I wasn't able to get my hands on a resistor this afternoon, but instead I hooked up another lab member to the GSR electrodes and compared what Actiview was showing me with what the .bdf file is recording.

When actiview was showing a value of 6500 nanosiemens, my bdf file gave the same value! I observed this when I opened my .bdf file in the EDF Browser and when I exported my signals to a text file. So it appears that the conversion is already applied to the data when you export it in some way, and all I would need to do is to convert the values from nanosiemens to microsiemens. This would explain the weird values I was seeing, as I was applying the conversion calculation twice.

Best wishes,

Sam

Post Reply