pspm_scr_pp filtering out all my data

[lucile26742]

Hello !

I am trying to use PsPM to clean my skin conductance data but the function is filtering out all my data even though it shouldn't (i.e., visual checks show the data is contained between 0.5 and 60 microsiemens, no clipping, no slopes greater than 10 seimens). I suspect there may be issues with the units but have not found a solution.

For information, my data comes from a Labchart matlab file, channel 2, is in siemens, (note that I've tried converting the data to microseimens prior to importing it with PsPM but this does not resolve the issue), and was downsampled to 32Hz (note that I've tried not downsampling it and I have the same issue). I've tried playing around with the filtering settings but even if I set extremely generous parameters, all the data still gets filtered out. I'm pasting my script here in the hopes that you may be able to help:

% Specify whether you want to overwrite files or not
options = struct('overwrite', 1, 'missing_epochs_filename', fullfile([missing_Epochs_Path id '.mat']));

% Specify import settings
import{1}.type = 'scr'; % Type of data
import{1}.channel = 2; % Which channel is the SCR data in
import{1}.transfer = 'none'; % Transfer function: since our setup directly measures conductance these are the settings to use
import{2}.type = 'marker'; % Extract triggers from labchart file as well

% Import
[sts, outfile] = pspm_import(this_Rawfile, 'labchartmat_in', import, options);

%% Trim data ----------------------------------------------------------

[sts, newdatafile] = pspm_trim(outfile, -trim_Envelope, trim_Envelope, 'marker', options);

%% Automated artefact detection ---------------------------------------

% Specify some settings
options.missing_epochs_filename = fullfile([missing_Epochs_Path id '.mat']);
options.min = 0.05;
options.max = 60;
options.slope = 10;
options.deflection_threshold = 0.1; 

[sts, missing_epochs_file] = pspm_scr_pp(newdatafile, options);

load(missing_epochs_file)

epochs_missing

epochs = epochs_missing;

save(missing_epochs_file, 'epochs');

Thank you so much for your time and attention!

FYI: The output of pspm_scr_pp is a variable called 'missing_epochs' but pspm_glm takes a variable called 'epochs', so I rename them in between.

[dominikbach]

Thanks for reporting this. The data need to be in mcS for the algorithm to use the correct cutoffs, but you have already tried that. Could you send us a test file?

[dominikbach]

Indeed, there was a bug in pspm_scr_pp that could filter out all data. This is now fixed in the develop branch. Furthermore, we made the function work more gracefully with different sample rates: the arguments clipping_window_size and clipping_step_size are now provided in seconds rather than number of data points.

One note of caution: all artefact detection thresholds are set in µS (physiological range: around 0-20). If the user provides data in different units, then the thresholds can be adapted via the function options.

[lucile26742]

Thank you so much !