test unmixing on spectral confocal image

[matfallet]

We are interested to test your algorithm on spectral confocal images (Zeiss LSM 980 NIR). Can we test your algorithm with TIFF lambda images (35 images) and do you need individual spectrum or do you evaluate them ? Thanks

[WANGRUOGU]

Thanks for your interest. May I ask your name or institution? Do you have reference images for these images?The spectral signatures or endmembers can be evaluated with reference images before unmixing. Otherwise, SL-PRU can be applied if the endmembers are known in advance.

Ruogu Wang

On Mon, Jul 24, 2023 at 8:02 AM matfallet @.***> wrote:

We are interested to test your algorithm on spectral confocal images (Zeiss LSM 980 NIR). Can we test your algorithm with TIFF lambda images (35 images) and do you need individual spectrum or do you evaluate them ? Thanks

— Reply to this email directly, view it on GitHub https://github.com/WANGRUOGU/SL-PRU/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALV422OYQQTGB6VSRFNGSSTXRZP6PANCNFSM6AAAAAA2VPGOCA . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[matfallet]

I am working in CNRS France in the Centre d'immunologie de Marseille Luminy. I am ingineer in microscopy and image processing. Yes I have reference image for each fluorochrome.

[WANGRUOGU]

Sorry for the late reply. You may extract the endmember from the reference images using PNMF.m, respectively. You can also skip this step if you have the spectral signitures. You may estimate the abundances using UnmixSpIm.m. You can follow the example in demo.m.

[matfallet]

Does-it works with TIF images ?

[JAMoreno-Larios]

Hi Mat,

You would need to load the TIF files yourself into MATLAB and arrange them into an array like the one used for test data. I think it was a N rows by C columns, where N is the number of pixels your image has and C is the number of channels.

Best, José

[WANGRUOGU]

Sure, I just uploaded the .m file called readMultipageTiff for reading tiff files.

[matfallet]

Ok great could you give me your mail in order I send you my image to test your algo, I am sorry I do not understand if I need to add the function readtiff in your demo.m ? Could you write the result tif as well in your Github ?

[matfallet]

Picasso code to unmix is very easy to use, you can maybe do the same to allow user to test your algo . I click the unmix code and just put before my image in the "data" folder , the runmix image is saved in result folder :

[WANGRUOGU]

Thanks for your advice. I modified my code and you may test it now.

[matfallet]

thanks, If I have no endmembers, can I test it ? Do I need to comment the line : Load endmembers.mat in the demo.m ? My mixture.tif contain 8 channels but I have 4 endmembers.

[WANGRUOGU]

SL-PRU needs the endmembers as prior information. You may comment the line -- load endmembers.mat in demo and load the 4 endmembers you used in your image.