FTIR Image data processing (correcting energy axis)

I'm working on processing some FTIR imaging data from an Agilent FTIR microscope with SMAK 3.0.7. The .dmt file loads into SMAK correctly and I see a new Data Channel labeled "FTIR" created along with the correctly processed mozaic tiling in the Display window. When I define the .hdf5 spectrum file (under MCA Spectra->Define MCA File) the x-axis untis are not displayed correctly. My source data should start at 898.67 cm^-1 and end at 3801.01 cm^-1, however it looks like the x-axis has loaded so that the first value is at 466 cm^-1 and the last data point is at 3468 cm^-1 (so the start position is off, as are the bins).

I can see there is an option in MCA Spectra->FTIR Support to load IR Settings, but it's looking for a .json file which I do not have and am unsure the formatting of (and am not sure this will fix the underlying issue).

Is there a recommended way to correct this or set the start value of the energy axis (particularly one that's not an integer, but somethign with several decimal places) as well as my bin size (should be 1.928467 cm-1/bin)?

Thanks!

7 comments

Comments (7)

MJ P

Oct 25, 2023

With the MCA View option set to Specturm I see the FTIR data plotted from 466.0 to 3468.623545 (copy/pasted to clipboard so I could confirm values). You're correct, it looks like the 1.928467 cm-1 spacing is being picked up correctly, but the start value of our spectrum should be 898.665750 cm-1 as far as I can tell. The real test here is where our peaks show up, and they're all shifted by the same amount (even the telltale CO2 peak is shifted too). The Agilent instrument starting range that we specify is 900-3900 cm-1 and appears ~~correct on the instrument itself. When loading our hdf5 files into the data processing package Orange (now named Quasar) it also shows this energy range, with a starting energy of 898.665750 and going up to 3901.289295 cm-1. The data from the Leica instrument at SSRL DOES display the correct energy scale, so this appears to be an Agilent-specific issue. The absolute energy for our Agilent data is off by 432.66575 cm-1 for each data point. Shifting the energy scale I have in SMAK right now by this amount would make 466.0-to-3468.623545 cm-1 into the expected 898.665750-to-3901.289295 cm-1 range. Thanks for any suggestions! ... luckily the Agilent data isn't the most pressing to process at the moment, as we've got lots of Leica data to get through ;)

Edited

samsxrays

Oct 25, 2023

Replying to

Ok - that's manageable at least. Not sure why the Agilent data file reader is off that way -- Its not overly documented, but I'll take a look and see if I can find anything in the data file that looks like an offset. If so, I'll make an edit in the reader. Otherwise, I can put an energy offset in SMAK for you -- would it be "better" to have the energy offset (1) entered in the first read of the FTIR data file so that it is a fix for the data file at first reading - but always a prompt when loading a data file for the first time, (2) a fix to the FTIR data file (i.e. MCA type data) after the initial read that gets saved in the data file, (3) an offset that you apply each time you open a FTIR (MCA) data file session?

samsxrays

Oct 25, 2023

Replying to

Ok - I found the error! It's quite stupid the way the data for the x-axis stored. There was an error in the file reader... regardless, it is fixed in the code now, the next release (3.0.8) will have the proper code for reading the wavenumbers correctly. You'll want to reload the raw data from the originals (ie. remove any of the -ftir.hdf5 files and reopen the original data file) to get the correct values.

MJ P

Oct 25, 2023

Replying to

Great looks like it's working as expected now! Thanks.

Problem solved!!

samsxrays

Oct 24, 2023

The MCA Spectra -> FTIR Support -> load IR Settings is for the background subtraction parameters, so won't adjust your FTIR x-axis.

I presume you've done the MCA Spectra -> MCA View Options -> and selected "Spectrum" over "Bins" (otherwise you probably wouldn't have seen the wavenumbers to start with!).

Looking at the Agilent loading code, the reader is extracting several binary bits for the spectrum, treating the first wavenumber as an integer, the number of points in the spectrum as an integer, and then the wavenumber step as a double precision float. I will look at the parameters in the file and see if we can find the correct interpretation of the spectrum from the data file...

Cheers!

MJ P

Oct 24, 2023

Replying to

Yes, Spectrum view option was set. If it's set to Bins then the energy axis starts at zero and gees up to 1557 in this case.

samsxrays

Oct 24, 2023

Replying to

Can you actually confirm that the peak positions in the spectra are NOT correct. The spacing in the (sample) dmt files that I have from you from an earlier file are at 1.928467 cm-1 (which you think is correct?). I would think that your data range that you are hoping the see is a slightly different range in the spectral array - maybe something that was set in the viewer of the Agilent? The data file is pretty "set" in those numbers, meaning there's not a good alternate interpretation of those binary numbers... Unless you think the instrument itself was calibrated poorly and you have a different calibration you want to override with?