search

ime-luebeck / ecg-removal

Algorithms and evaluation toolkit for removing strong cardiac interference from surface EMG measurements
Other
8 stars 5 forks source link

Discrepancy in filter_signals #1

Closed GallVp closed 1 year ago

GallVp commented 1 year ago

The help section says (sigIn{ii, chan}(:, 4)):

% passTime: (Optional) If this is 1, the signal sigIn{ii, chan}(:, 4) is passed as an additional % input vector to the filter function, generally assuming that this contains the measurement time % in ms. If 0 (default), no additional vector is passed.

Where as the code uses (sigIn{ii, 4}):

[sigsOutLoc{:}] = algo(sigIn{ii, chan}(:, 1), sigIn{ii, 4});

I think the time vector should not be placed at sigIn{ii, 4} because it either limits the number of signal channels to 3 or creates the problem of non-contiguous signal channels in sigIn cell array. Rather, it should be placed at sigIn{ii, chan}(:, 2) or sigIn{ii, chan}(:, 3) depending on whether auxChannel and passTime are true or false.

Cheers Usman R.

GallVp commented 1 year ago

Good job on the toolbox, btw. Thank you for your efforts.

GallVp commented 1 year ago

fs = 512; on line 72 in ecg_removal masks fs = 1024; on line 56. DOn't you think that this making can result in wrong processing of sigIn{1, jj} and sigIn{2, jj} in the remaining steps.

I think the fs = 512 should be declared at the top as:

fs = 1024; fs_pressue = 512;

e-pet commented 1 year ago

Hi Usman,

thank you so much for the detailed reports, and sorry for my ridiculously slow response time!

Both issues were very well spotted and should be fixed in the newest version. :-)

(I modified filter_signals to now allow the user to specify where the extra channels are to be found. That should provide sufficient flexibility for all use cases, I hope. The sampling rate issue was a bug I apparently introduced a few months ago while working on another bug fix related to different sampling rates.)