Noise removal from mass profile using reference mass profile

I have imported mass spectra from two mzXML files (noise, pure) , which work fine. Associated functions mspeaks, msdotplot, all work perfectly fine. I used mspeaks to get the most relevant peaks, and using 'Children','get' functions, I got the XData, YData, ZData, which are the M/Z values, Retentime time values, and intensity values respectively for each sample.

I stiched the X,Y,Z Data for each sample, noise and pure to get cubes.

'pure' is a sample which contains my 'component of interest' + 'solvent noise' of size (206842,3)

'noise' is a sample which contains the 'solvent noise' of size (196452,3)

This is the code used to generate the mass profile for 'pure'

hp = plot3(pure(:,1),pure(:,2),pure(:,3),'r.','DisplayName','Centroids','Tag','msdotplot');
set(hp,'MarkerSize',1),
ylim([903 1626]) %% set time
view(0,90),white,box on, title('pure')
xlabel('M/Z'),ylabel('R time (min)')
xlim([400 1300]) %% set M/Z

Similarly, the profile for noise can be plotted :-

When I overlap the files, noise (black) and pure (red), they look like this :-

Is there a simple way to filter out the noise (black) from pure (red) and get a corrected 'pure sample'? I have uploaded the files as a .mat (with x = M/Z, y = time (min), z = intensity)

[darova]

Did you try to binarize images and substract?

[Sindhuraj Mukherjee]

Yes. However, the image sizes are different even though they look they are being plotted on the same scales. I understand you are suggesting I try this through the image processing perspective, but that wont work.

'pure' is a sample which contains my 'component of interest' + 'solvent noise' of size (206842,3)

'noise' is a sample which contains the 'solvent noise' of size (196452,3)

Simple, subtraction wont work anyways. The x-axis is M/Z which is different for different samples, the retention time which is the y axis is actually also shifted. And as mentioned, cube sizes are different.

Thanks for your suggestion though, you are free to try this, I have attached the mat file with the two cubes in the question.

[darova]

I'm not good at signals, Maybe you need something like fft filtering