Ratio and Background substraction operations
Locked
Ratio and Background substraction operations
 
|
My question is more mathematical or theoretical: Whether the detector is logarithmic or linear in output vs. light intensity, is it correct to do a division to get a ratio image in both cases? Or should it be a subtraction (?!) considering that to remove the background from an image can be by subtracting (if logarithmic detector) or dividing (if linear detector)?
Monique Vasseur Département de biochimie Université de Montréal |
Re: Ratio and Background subtraction operations
|
|
I think that you are right in the sense that it
depends on the detector. In case of MRI for example the bias field is multiplicative, therefore one must divide or subtract after log transform. In the case of microscopy, it depends on your acquisition camera. One way to test it, is to image a phantom with two areas with a sharp contrast, and image a flat background. By looking at some profiles you should be able to figure it out. ___________________________ Olivier Salvado Biomedical Engineering Department Case Western Reserve University -----Original Message----- From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Vasseur Monique Sent: Thursday, January 12, 2006 10:26 AM To: [hidden email] Subject: Ratio and Background substraction operations My question is more mathematical or theoretical: Whether the detector is logarithmic or linear in output vs. light intensity, is it correct to do a division to get a ratio image in both cases? Or should it be a subtraction (?!) considering that to remove the background from an image can be by subtracting (if logarithmic detector) or dividing (if linear detector)? Monique Vasseur Département de biochimie Université de Montréal |
Re: Ratio and Background subtraction operations
|
|
With fluorescence where the intensity is directly proportional to the
number of molecules and the detector is linear (e.g. CCD camera), then division after flatfield correction (if necessary) and background subtraction is the correct method. > I think that you are right in the sense that it > depends on the detector. In case of MRI for > example the bias field is multiplicative, > therefore one must divide or subtract after log > transform. In the case of microscopy, it depends > on your acquisition camera. > > One way to test it, is to image a phantom with two > areas with a sharp contrast, and image a flat > background. By looking at some profiles you should > be able to figure it out. > > ___________________________ > > Olivier Salvado > Biomedical Engineering Department > Case Western Reserve University > > -----Original Message----- > From: ImageJ Interest Group > [mailto:[hidden email]] On Behalf Of Vasseur > Monique > Sent: Thursday, January 12, 2006 10:26 AM > To: [hidden email] > Subject: Ratio and Background substraction > operations > > My question is more mathematical or theoretical: > Whether the detector is logarithmic or linear in > output vs. light intensity, is it correct to do a > division to get a ratio image in both cases? Or > should it be a subtraction (?!) considering that > to remove the background from an image can be by > subtracting (if logarithmic detector) or dividing > (if linear detector)? > > > > Monique Vasseur > > Département de biochimie > > Université de Montréal > > > > > _________________________________________ Michael Cammer Analytical Imaging Facility and Dept. ASB Biophotonics Innovation Laboratory Albert Einstein College of Medicine 1300 Morris Park Avenue, Bronx, NY 10461 718-430-2890 Fax 718-430-8996 work: http://www.aecom.yu.edu/aif/ personal: http://coxcammer.com/ |
Re: Ratio and Background subtraction operations
|
|
Michael Cammer a écrit :
> With fluorescence where the intensity is directly proportional to the > number of molecules and the detector is linear (e.g. CCD camera), then > division after flatfield correction (if necessary) and background > subtraction is the correct method. > Hello, Just as a remark, in biology applied microscopy, fluorescence is not so proportional to the number of molecules. One knows that environment of the dye could have a dramatic inluence on quantum efficiency (presence of quenchers, pH differences inducing spectral modifications,... in the different cell compartments). I agree for all the other considerations ;-) > > >>I think that you are right in the sense that it >>depends on the detector. In case of MRI for >>example the bias field is multiplicative, >>therefore one must divide or subtract after log >>transform. In the case of microscopy, it depends >>on your acquisition camera. >> >>One way to test it, is to image a phantom with two >>areas with a sharp contrast, and image a flat >>background. By looking at some profiles you should >>be able to figure it out. >> >>___________________________ >> >>Olivier Salvado >>Biomedical Engineering Department >>Case Western Reserve University >> >>-----Original Message----- >>From: ImageJ Interest Group >>[mailto:[hidden email]] On Behalf Of Vasseur >>Monique >>Sent: Thursday, January 12, 2006 10:26 AM >>To: [hidden email] >>Subject: Ratio and Background substraction >>operations >> >>My question is more mathematical or theoretical: >>Whether the detector is logarithmic or linear in >>output vs. light intensity, is it correct to do a >>division to get a ratio image in both cases? Or >>should it be a subtraction (?!) considering that >>to remove the background from an image can be by >>subtracting (if logarithmic detector) or dividing >>(if linear detector)? >> >> >> >>Monique Vasseur >> >>Département de biochimie >> >>Université de Montréal >> >> >> >> >> > > > > _________________________________________ > Michael Cammer > Analytical Imaging Facility and > Dept. ASB Biophotonics Innovation Laboratory > Albert Einstein College of Medicine > 1300 Morris Park Avenue, Bronx, NY 10461 > 718-430-2890 Fax 718-430-8996 > work: http://www.aecom.yu.edu/aif/ > personal: http://coxcammer.com/ > -- CHAMOT Christophe --------------------------------------------------------------------- Plate-Forme de Recherche IFR117 "Imageries des Processus Dynamiques en Biologie Cellulaire et Biologie du Développement " Institut Jacques Monod, CNRS, Universités Paris 6 et 7 2, place Jussieu - Tour 43 75251 Paris cedex 05 Tel: 01 44 27 57 84 http://www.ijm.jussieu.fr/ --------------------------------------------------------------------- |
Re: Ratio and Background subtraction operations
|
|
In reply to this post by Michael Cammer
On Tuesday 17 January 2006 01:28, Michael Cammer wrote:
> With fluorescence where the intensity is directly proportional to the > number of molecules and the detector is linear (e.g. CCD camera), then > division after flatfield correction (if necessary) and background > subtraction is the correct method. Division by the background is used to estimate the light absorption by the sample in bright microscopy by looking at the ratio of light transmission (ie (sample/brightfield)*greyscale_range What is the rational for using it in darkfield images? There should be no background lighting, so what is the denominator? Doesn't subtraction of the background suffice? Cheers, Gabriel |
Re: Ratio and Background subtraction operations
|
|
In reply to this post by Olivier Salvado
In my experience the lineraity depends strongly on the
CCD camera. If the camera has non-linear filtering capabilites then the relationship comes out also as non linear. The other thing you should have in mind is the range of acquisition. In the grayscale you map only between 0 and 255, so anything that has too low or too high intensity goes to 0 and 255, respectively. In order to be shure you should do calibration. Cheers D. Prodanov ------------------------------ Date: Tue, 17 Jan 2006 10:02:15 +0100 From: CHAMOT Christophe <[hidden email]> Subject: Re: Ratio and Background subtraction operations Michael Cammer a écrit : > With fluorescence where the intensity is directly proportional to the > number of molecules and the detector is linear (e.g. CCD camera), then > division after flatfield correction (if necessary) and background > subtraction is the correct method. > Hello, Just as a remark, in biology applied microscopy, fluorescence is not so proportional to the number of molecules. One knows that environment of the dye could have a dramatic inluence on quantum efficiency (presence of quenchers, pH differences inducing spectral modifications,... in the different cell compartments). I agree for all the other considerations ;-) > > >>I think that you are right in the sense that it >>depends on the detector. In case of MRI for >>example the bias field is multiplicative, >>therefore one must divide or subtract after log >>transform. In the case of microscopy, it depends >>on your acquisition camera. >> >>One way to test it, is to image a phantom with two >>areas with a sharp contrast, and image a flat >>background. By looking at some profiles you should >>be able to figure it out. >> >>___________________________ >> >>Olivier Salvado >>Biomedical Engineering Department >>Case Western Reserve University >> >>-----Original Message----- >>From: ImageJ Interest Group >>[mailto:[hidden email]] On Behalf Of Vasseur >>Monique >>Sent: Thursday, January 12, 2006 10:26 AM >>To: [hidden email] >>Subject: Ratio and Background substraction >>operations >> >>My question is more mathematical or theoretical: >>Whether the detector is logarithmic or linear in >>output vs. light intensity, is it correct to do a >>division to get a ratio image in both cases? Or >>should it be a subtraction (?!) considering that >>to remove the background from an image can be by >>subtracting (if logarithmic detector) or dividing >>(if linear detector)? >> >> >> >>Monique Vasseur >> >>Département de biochimie >> >>Université de Montréal >> >> >> >> >> > > > > _________________________________________ > Michael Cammer > Analytical Imaging Facility and > Dept. ASB Biophotonics Innovation Laboratory > Albert Einstein College of Medicine > 1300 Morris Park Avenue, Bronx, NY 10461 > 718-430-2890 Fax 718-430-8996 > work: http://www.aecom.yu.edu/aif/ > personal: http://coxcammer.com/ > -- CHAMOT Christophe --------------------------------------------------------------------- Plate-Forme de Recherche IFR117 "Imageries des Processus Dynamiques en Biologie Cellulaire et Biologie du Développement " Institut Jacques Monod, CNRS, Universités Paris 6 et 7 2, place Jussieu - Tour 43 75251 Paris cedex 05 Tel: 01 44 27 57 84 http://www.ijm.jussieu.fr/ --------------------------------------------------------------------- ------------------------------ Date: Tue, 17 Jan 2006 09:20:34 +0000 From: Gabriel Landini <[hidden email]> Subject: Re: Ratio and Background subtraction operations On Tuesday 17 January 2006 01:28, Michael Cammer wrote: > With fluorescence where the intensity is directly proportional to the > number of molecules and the detector is linear (e.g. CCD camera), then > division after flatfield correction (if necessary) and background > subtraction is the correct method. Division by the background is used to estimate the light absorption by the sample in bright microscopy by looking at the ratio of light transmission (ie (sample/brightfield)*greyscale_range What is the rational for using it in darkfield images? There should be no background lighting, so what is the denominator? Doesn't subtraction of the background suffice? Cheers, Gabriel ------------------------------ Date: Tue, 17 Jan 2006 09:40:07 -0000 From: "Wang, Junsheng" <[hidden email]> Subject: Background subtraction operations As to the uneven illuminative image stack from X-ray transmission, it is difficult to substract a proper backgroud per image. I'm planning to do the z-direction median filter and trying to generate a backgroud from the stack. Can anybody point me to a plugin which can do this? Cheers, Junsheng Wang ------------------------------ |
Re: Ratio and Background subtraction operations
|
|
Everything is for the best in the best of all possible worlds?
Hey, we're just talking intensities; try measuring lifetimes! > Hello, > Just as a remark, in biology applied microscopy, fluorescence is not so > proportional to the number of molecules. One knows that environment of > the dye could have a dramatic inluence on quantum efficiency (presence > of quenchers, pH differences inducing spectral modifications,... in the > different cell compartments). > I agree for all the other considerations ;-) |
Re: Ratio and Background subtraction operations
|
|
In reply to this post by Olivier Salvado
-----Message d'origine-----
De : ImageJ Interest Group [mailto:[hidden email]] De la part de Gabriel Landini Envoyé : 17 janvier 2006 04:21 À : [hidden email] Objet : Re: Ratio and Background subtraction operations On Tuesday 17 January 2006 01:28, Michael Cammer wrote: > With fluorescence where the intensity is directly proportional to the > number of molecules and the detector is linear (e.g. CCD camera), then > division after flatfield correction (if necessary) and background > subtraction is the correct method. Division by the background is used to estimate the light absorption by the sample in bright microscopy by looking at the ratio of light transmission (ie (sample/brightfield)*greyscale_range What is the rational for using it in darkfield images? There should be no background lighting, so what is the denominator? Doesn't subtraction of the background suffice? Cheers, Gabriel Ratios and background corrections are used in FRET experiments where there is bleedthrough of donor-fluorophore1 over the excitation and emission spectra of acceptor-fluorophore2. So I was wondering if I should have a different calculation approach depending if my pictures where coming from the confocal microscope (PMT detector - so logarithmic one) and fluorescence microscope equipped with a CCD camera (linear detector)? Monique |
«
Return to ImageJ
|
1 view|%1 views
| Free forum by Nabble | Edit this page |