spectral unmixing without "clean" reference signals?

[Guenter Giese]

Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

(A previous version of this contribution has been also sent to the ImageJ
list)

Dear list members,

how to do spectral unmixing without "clean" reference signals? I have stacks
of images recorded with a ccd camera at different excitation / emission
wavelength combinations. Besides specific fluorescence signal, there is
considerable autofluorescence signal as well as some crosstalk
(bleedthrough) of fluorophores.

Unfortunately, I have no region with single fluorescence signal to be used
as a reference, and it would be impossible or very time-consuming for
several reasons (different tissue types, ages, species, tissue preparation
methods etc.) to prepare individual reference probes.

Of course there are regions in different stacks in which any type of
fluorescence is prominent (but not isolated), and these regions could be
used as reference regions.

Is there a way to determine, without "clean" reference signals, the matrix
values to be used with, e.g., the ImageJ Spectral Unmixing plugin for from
J. Walter, or for general spectral unmixing purposes?

The solution I would favour would be interactive (using sliders / updated
result windows).



Best,

Guenter Giese

------------------------------------------
Dr. Guenter Giese
Light Microscopy Facility Manager
Dept. of Biomedical Optics
MPI fuer Medizinische Forschung Jahnstr. 29
D-69120 Heidelberg, Germany
Phone [phone] (Fax: -325)
e-mail: [hidden email]

[Ignatius, Mike]

Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

Not sure if this helps but some folks like these.  I realize folks are
trying to do without, but this may confirm the accuracy of whatever
approach has been used.

Probes/Invitrogen FocalCheck(tm) fluorescence microscope test slide #2
*for spectral imaging systems*  SKU# F36913   Lots of reference colors
to parse out, all mounted on one slide.  

        ex/em
Ring 503/511 nm
Core 511/524 nm
Ring 541/555 nm
Core 545/565 nm
Ring 589/613 nm
Core 578/605 nm
Ring 671/692 nm
Core 657/676 nm

Bob Zucker featured them in this article.

Cytometry A. 2007 Mar;71(3):174-89. Links
Reliability of confocal microscopy spectral imaging systems: use of
multispectral beads.Zucker RM, Rigby P, Clements I, Salmon W, Chua M.
Reproductive Toxicology Division (MD-67), National Health and
Environmental Effects Research Laboratory, Office of Research and
Development, U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina 27711, USA. zucker.robert@.epa.gov

Mike Ignatius
Molecular Probes/Invitrogen

-----Original Message-----
From: Confocal Microscopy List [mailto:[hidden email]] On
Behalf Of Guenter Giese
Sent: Monday, May 05, 2008 9:36 AM
To: [hidden email]
Subject: spectral unmixing without "clean" reference signals?

Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

(A previous version of this contribution has been also sent to the
ImageJ
list)

Dear list members,

how to do spectral unmixing without "clean" reference signals? I have
stacks
of images recorded with a ccd camera at different excitation / emission
wavelength combinations. Besides specific fluorescence signal, there is
considerable autofluorescence signal as well as some crosstalk
(bleedthrough) of fluorophores.

Unfortunately, I have no region with single fluorescence signal to be
used
as a reference, and it would be impossible or very time-consuming for
several reasons (different tissue types, ages, species, tissue
preparation
methods etc.) to prepare individual reference probes.

Of course there are regions in different stacks in which any type of
fluorescence is prominent (but not isolated), and these regions could be
used as reference regions.

Is there a way to determine, without "clean" reference signals, the
matrix
values to be used with, e.g., the ImageJ Spectral Unmixing plugin for
from
J. Walter, or for general spectral unmixing purposes?

The solution I would favour would be interactive (using sliders /
updated
result windows).



Best,

Guenter Giese

------------------------------------------
Dr. Guenter Giese
Light Microscopy Facility Manager
Dept. of Biomedical Optics
MPI fuer Medizinische Forschung Jahnstr. 29
D-69120 Heidelberg, Germany
Phone [phone] (Fax: -325)
e-mail: [hidden email]

[jens rietdorf]

Dear Guenter,

it is very difficult to predict/fit the modulation of the spectral signal of a given fluorophore by your particular detection device. A good estimate can typically be derived from a reference measurement of the fluorophores on the same detection device with low 'background' sample, eg. cultured fibroblast cells. The 'background' estimation is of course best done on your actual sample.

Hope that helps,

Guenter Giese wrote

how to do spectral unmixing without "clean" reference signals?

Unfortunately, I have no region with single fluorescence signal to be used
as a reference, and it would be impossible or very time-consuming
...to prepare individual reference probes.

Is there a way to determine, without "clean" reference signals, the matrix
values to be used with, e.g., the ImageJ Spectral Unmixing plugin for from
J. Walter, or for general spectral unmixing purposes?

------------------------------------------
Dr. Guenter Giese
Light Microscopy Facility Manager
Dept. of Biomedical Optics
MPI fuer Medizinische Forschung Jahnstr. 29
D-69120 Heidelberg, Germany
Phone [phone] (Fax: -325)
e-mail: [e-mail]