Importing from Thermo Noran EDS Spectrum Images

[Luiz Fernando Zagonel]

Dear Hyperspy users, 

On the TEM available here, I can only acquire nice EDX (EDS) maps using a proprietary software called Thermo Fisher (Scientifc) "Noran System Six", also called NSS. Even if this is the best acquisition method, I did not manage to export the data to any other format using the acquisition (NSS) software. Also, I could find no software that could open the spectrum images saved using this software.

One example of Spectrum Images is available on the links below:

https://dl.dropboxusercontent.com/u/62740343/Base.si

https://dl.dropboxusercontent.com/u/62740343/Base.siref

Apparently, the "siref" file corresponds to the header in a TIFF-like format and the "si" file has all the data in TIFF-like format. At least, it says so in the Thermo Norar System Six documentation. Still, I could not read it using any software I know (like ImageJ or DigitalMicrograph). Opening as 'raw data' also did not work. 

I would really liked to explore this data using Hyperspy and go beyond the acquisition software limitations. 

Any help is welcome. 

With best regards,

Luiz Fernando Zagonel.

[Gavin Mitchson]

Hello Luiz and other Hyperspy users,

I am running into the same problem Luiz describes. I would also very much like to explore my data beyond the capabilities in NSS using Hyperspy. Does anyone have any thoughts or suggestions?

Best regards,

Gavin 

[Thomas Aarholt]

I'll be looking into this one over the next couple of weeks/Christmas. We have a NSS EDX as well, and it would be nice to use Hyperspy on it.

Not sure why the previous two emails are listed as "deleted" on the Google Groups?

- Tom

[Thomas Aarholt]

This has come up again here in Oxford. Has anyone found any solutions to exporting NSS into a useful format, ideally readable by Hyperspy?

- Tom

[Thomas Aarholt]

I thought I'd add a bit of info here. On our NSS system, an EDX "site" contains four files with the following extensions:

".sitif", ".siref", ".csi" and ".si".

The first two, I've discovered just now, are just normal tifs, with no interesting header info (afaik):

Ru nning the following on either of the two tifs produces just a few random numbers of header info, but does return an image. 

"

from PIL import Image

im = Image.open("Base(3).sitif")

for tag in im.tag:

    print(tag)

"

The reason the siref file is larger is because it is an rgba tif, whilst the other is just greyscale.

Both of these can be opened in an image browser just by changing the extension of the file to ".tif".

They are not the same image. A slight movement in the position of the features on the image (i.e. drift) leads me to believe that one is acquired initially and the other after some time has passed.

I have found no way to extract any information, even calibration information (like the energy range selected for the EDX acquisition) other than using the Thermo NSS software.

Of the remaining two files, the csi file is useless, and the si file contains the entire dataset, which we still don't know the spec of, afaik.

- Tom

[Pierre Burdet]

Dear Hyperspy users,

Good news a solution exists. Bad news you have to pay for it. Thermo is providing an option to export NSS data in an open format. I don't know the details. The best is to contact your NSS provider.

Best regards

Pierre Burdet

[Pierre Burdet]

Dear Hyperspy users,

 

The solution was just there. Thermo provides a software to convert their format (.csi) into rpl format (.rpl). The program is call "SIConvert.exe" and can be found in "C:\Program Files (x86)\Thermo Scientific\NSS".

 

All metadata are loss. They can be retreived using the program called "SIAttributes.exe" in the same folder and have to be manually added in hyperspy. For example:  

 

>>> s = hs.load('Map.rpl')

>>> s.set_signal_type('EDS_SEM')

>>>names = ['x', 'y','Energy']

>>>units = ['${\mu}m$','${\mu}m$','keV']

>>>scales = [0.466, 0.466, 0.01]

>>>for i in range(3):

>>>    ax = s.axes_manager[i]

>>>    ax.name = names[i]

>>>    ax.units = units[i]

>>>    ax.scale = scales[i] 

>>> elements = ['Ti', 'Cu']

>>> s.set_elements(elements)

>>> s.set_microscope_parameters(beam_energy=10)

[Richard Webster]

Not wanting to bring up a potentially dead topic, but I have been looking into this option recently and as this is the most prominent result from a google search I thought I would try and add to Pierre's info.

I have found that the .rpl file created by "SIConvert.exe" is missing some information and should have the following tags appended to it before I can import into hyperspy without any issues.

width-scale    0.0197606
width-origin    0.0
width-name     nm
width-units    nm

height-scale    0.0197606
height-origin    0.0
height-name     nm
height-units    nm

depth-origin    0.0
depth-scale    0.01
depth-units    keV

(the values here were found using the "SIAttributes.exe" or in the "extracted spectrum.emsa" file)

NB I'm not sure what the height-name value is for, so I just put the units there.

Doing this also seemed to work with importing noran rpl data into digital micrograph using Luiz Fernando Zagonel's script found here: https://hyperspy.readthedocs.io/en/stable/user_guide/io.html#import-rpl