How to correct background to extract fine features of a signal containing multiple gaussian envelopes?

I am trying to extract fine features and plot them from a signal that consists of two distinct gaussian envelopes. I used the background correction code below https://www.mathworks.com/matlabcentral/fileexchange/27429-background-correction

Apparently, the code works only for signals with one gaussian envelope (I may be wrong?); but i could not get the desired result. I cannot simply apply gaussian filtering because it will modify the linewidths of the fine features, which are of key importance in this singal.

Any help to resolve this issue would be highly appreciated. I have attached a .mat file to give folks an idea about how the vectors look like.

Answer 1

[Image Analyst]

What is this? What generated the signal? It already looks like a baseline-corrected signal. I'm not sure you'd want to do anything to that signal.

s = load('ct8.mat')

s = struct with fields:
    x1: [1340×1 double]
    y1: [1340×1 double]

x = s.x1;
y = s.y1;
plot(x, y, 'b-');
grid on;

figure;
s = load('ct8.mat');
x = s.x1;
y = s.y1;
subplot(2, 1, 1);
plot(x, y, 'b-');
grid on;
smoothy = sgolayfilt(y, 2, 151);
hold on;
plot(x, smoothy, 'r-', 'LineWidth', 2)
subplot(2, 1, 2);
yDiff = y - smoothy;
plot(x, yDiff, 'b-');
grid on;

If you were to subtract out the baseline, where would you put it? Can you attach a picture of the baseline drawn in red underneath the signal? By the way, I'm attaching some demos for you to adapt.

Or maybe you could use sgolayfilt to get a smoothed version of the signal and then subtract it.

[ct8]

Thanks @Image Analyst for your response! The correction you suggested works quite well. However, the expected "noise" is supposed to look like a group of extremely narrow lines with zero as baseline (since negative counts have no meaning in this specific signal). A typical example is attached here:

https://www.nature.com/articles/srep06302/figures/3 (Figure 3)

I hope this makes sense.

[Image Analyst]

I have other baseline correction methods. If I get time this weekend I can try them on your data. I don't have time today, probably.

[ct8]

Great no rush and thanks for the help!