Hello everybody !
I have a question about the "Gel analysis"-function of ImageJ: How does these function work ? What ist the scale of the y-axis and what is the unity of these axis or which values are used for the y-axis ? Is it just the gray values ? The Background ist, that I am working on the quantification of TLC and HPTLC with ImageJ. My idea is, that the gray value or the RGB-values of a TLC-scan, made by an usual office flatbed scanner, can be used as a detector signal, so that the ammount of substance must be propoprational to the peak-area, so that the TLC can be quantified like the gaschromatography or the HPLC. For this I am using the "Gel-analysis"-function of ImageJ. To test these method, I have dabbed on a dilution series of Glycin (0,01 mg/L - 1 g/L). I derivated the Glycin with Ninhydrin, scanned the TLC-plate and appraised the Peak-Area of the different concentrations. I was a bit surprised, because I got a logharitmical calibration curve - I expected a linear regression. C.B. |
On Saturday 28 Apr 2012 09:44:33 Christian Beringer wrote:
> My idea is, that the gray value or the RGB-values of a > TLC-scan, made by an usual office flatbed scanner, can be used as > a detector signal, so that the ammount of substance must be propoprational > to the peak-area, so that the TLC can be quantified like the > gaschromatography or the HPLC. Is that the right way of doing it? Scanners work by capturing the reflection of the light from the object. Can you get an measurement that should be obtained via transmittance? Or are you using a light box? Cheers Gabriel |
Hola Gabriel,
Relatively cheap desktop scanners are good for acquiring images, even at resolutions of approx. some 10 um and in transmission mode. Christian, but why do you acquire RGBs, are you analyzing each channel separately? Gary. On Apr 28, 2012, at 4:20 PM, Gabriel Landini wrote: > On Saturday 28 Apr 2012 09:44:33 Christian Beringer wrote: >> My idea is, that the gray value or the RGB-values of a >> TLC-scan, made by an usual office flatbed scanner, can be used as >> a detector signal, so that the ammount of substance must be propoprational >> to the peak-area, so that the TLC can be quantified like the >> gaschromatography or the HPLC. > > Is that the right way of doing it? > Scanners work by capturing the reflection of the light from the object. > Can you get an measurement that should be obtained via transmittance? Or are > you using a light box? > > Cheers > Gabriel |
In reply to this post by Gabriel Landini
Right now I am testing these method. I know, that flatbed-scanners work by
capturing the light which is reflected by the TLC-plate. But my consideration is, that more colour-pigments (which means more molecules of the substance that should be detected) also means less reflection. It is the same thing like scanning a text-document: The ink of the letters absorb the light, so less light is reflected. The relection of scattering of the light by the silica-gel of the TLC-plate is neglected, because it should be an systematic mistake. First tests with a dilution series of Glycin an derviatisation with Ninhydrin came to a logarithmic calibration-curve. So it seems like smaller amounts of substance really result in smaller peak-areas, but the calibration is not linear. But you can even see differences between the deep of the colours between the different concentrations. So I am now searching for the reason of these measurements and so I asked myself how the "Gel-Analysis" function works. Maybe it does not have a global scale and it just displays the grey values of the individual pixels against the highest grey value or anything else ... I thought, that the function is using a global scale like zero beeing bright white and 100 beeing fully black or something like this ... What is the difference between the "Gel analysis"-function an the function "Plot Profile" ? Greets C.B. 2012/4/28 Gabriel Landini <[hidden email]> > On Saturday 28 Apr 2012 09:44:33 Christian Beringer wrote: > > My idea is, that the gray value or the RGB-values of a > > TLC-scan, made by an usual office flatbed scanner, can be used as > > a detector signal, so that the ammount of substance must be > propoprational > > to the peak-area, so that the TLC can be quantified like the > > gaschromatography or the HPLC. > > Is that the right way of doing it? > Scanners work by capturing the reflection of the light from the object. > Can you get an measurement that should be obtained via transmittance? Or > are > you using a light box? > > Cheers > Gabriel > -- Mit freundlichen Grüßen - Christian Beringer - |
On Saturday 28 Apr 2012 16:16:23 Christian Beringer wrote:
> It is the same thing like scanning a text-document: > The ink of the letters absorb the light, so less light is reflected. I do not think the comparison is valid. Gel analysis is based on absorbance to estimate a quantity of light absorbing material. To measure absorbance, you need a known bright field intensity and use Beer-Lambert law: http://en.wikipedia.org/wiki/Beer%E2%80%93Lambert_law ) That is how optical density measurements are done in spectrophotometers. What you suggest is different: light goes through the light absorbant material, the fraction that remains after absorption and scattering bounces (reflects) on the white surface of the scanner lid, then goes back through the substance, it is absorbed and scattered for a second time, then it is captured by the sensor. This is not the same thing as scanning a flat, opaque document. Sure you will measure something, it won't be in optical density units as expected. The standard way of measuring this is using the ratio of the light passing through the material and the incident light, not the amount of reflected light. Some scanners have a light box attachment to scan 35mm slides and film. You probably need one of those to capture bright field images of your gels. That being said, I do not know how accurate the sensors are in office scanners. In addition many scanners have a very short depth of focus, which would not be ideal when scanning a gel that has a finite thickness. Hope this helps Gabriel |
I am not sure I agree with Gabriel. If you can develop a standard curve with known amounts, then I don't see how it matters what the methodology is for measuring the change in material or whether the fit is linear or logarithmic. You should be able to fit your unknowns to the standard curve and measure a relative amount of the material. Dave
On Apr 28, 2012, at 5:35 PM, Gabriel Landini wrote: > On Saturday 28 Apr 2012 16:16:23 Christian Beringer wrote: >> It is the same thing like scanning a text-document: >> The ink of the letters absorb the light, so less light is reflected. > > I do not think the comparison is valid. Gel analysis is based on absorbance to > estimate a quantity of light absorbing material. To measure absorbance, you > need a known bright field intensity and use Beer-Lambert law: > http://en.wikipedia.org/wiki/Beer%E2%80%93Lambert_law ) > That is how optical density measurements are done in spectrophotometers. > > What you suggest is different: light goes through the light absorbant > material, the fraction that remains after absorption and scattering bounces > (reflects) on the white surface of the scanner lid, then goes back through the > substance, it is absorbed and scattered for a second time, then it is captured > by the sensor. This is not the same thing as scanning a flat, opaque document. > > Sure you will measure something, it won't be in optical density units as > expected. The standard way of measuring this is using the ratio of the light > passing through the material and the incident light, not the amount of > reflected light. > > Some scanners have a light box attachment to scan 35mm slides and film. You > probably need one of those to capture bright field images of your gels. That > being said, I do not know how accurate the sensors are in office scanners. > In addition many scanners have a very short depth of focus, which would not be > ideal when scanning a gel that has a finite thickness. > > Hope this helps > Gabriel Visiting Professor David Knecht Beatson Institute for Cancer Research University of Glasgow Switchback Road, Bearsden Glasgow Scotland G61 1BD UK |
In reply to this post by Christian Beringer
>That
> being said, I do not know how accurate the sensors are in office scanners. > In addition many scanners have a very short depth of focus, which would not be > ideal when scanning a gel that has a finite thickness. According to some internet research the scanner sensor seems to be the same which are used for photometry ! The resolution is just limited by the CPU frequency and the amount of RAM of the used computer. I am not measuring electrophoretic gels, I want to use the Gel-Analysis-function for quantification of TLC. So the thickness doesn´t matter, because I am using HPTLC-plates. The silicagel-layer of these plates is as thin as a a sheet of paper and it is based on an aluminium-foil. My only problem is the scattering of the silica-gel. >Christian, but why do you acquire RGBs, are you analyzing each channel separately I think ImageJ allows also to analyse each channel separatly, but until now I just used the RGB-pictures (.tiff) and the Gel-Analysis-function. I have no idea how these function works with RGB-scans, but as I tested it with some .tiff made by Gimp it seems that the colour doesn´t matter for these function. But tomorrow I will use 8-bit pictures to measure the grey value. The crazy thing is, that the normal TLC-plates (thicker silica particels) work even better than the HPTLC-plates, I think that belongs to the scattering, which is increased by smaller particels, which are used for HPTLC-plates. |
In reply to this post by Knecht, David
I must agree with Gabriel. His analysis is right on target, since the
stain on a gel, autoradiogram, or tlc plate is actually a three dimensional object, having some significant thickness. The thickness can vary with concentration as well. Not only that, but the spots tend to be different sizes as well, as a function of concentration. As a result, the measurement that you get is a complex mixture of absorption, reflection, and reabsorption, and area-dependent. We have actually run the comparison, and it is difficult to set up a valid standard curve with conventional scanners. I suspect the we are also running into saturation problems as well. Joel On Sat, Apr 28, 2012 at 12:58 PM, David Knecht <[hidden email]>wrote: > I am not sure I agree with Gabriel. If you can develop a standard curve > with known amounts, then I don't see how it matters what the methodology is > for measuring the change in material or whether the fit is linear or > logarithmic. You should be able to fit your unknowns to the standard curve > and measure a relative amount of the material. Dave > > On Apr 28, 2012, at 5:35 PM, Gabriel Landini wrote: > > > On Saturday 28 Apr 2012 16:16:23 Christian Beringer wrote: > >> It is the same thing like scanning a text-document: > >> The ink of the letters absorb the light, so less light is reflected. > > > > I do not think the comparison is valid. Gel analysis is based on > absorbance to > > estimate a quantity of light absorbing material. To measure absorbance, > you > > need a known bright field intensity and use Beer-Lambert law: > > http://en.wikipedia.org/wiki/Beer%E2%80%93Lambert_law ) > > That is how optical density measurements are done in spectrophotometers. > > > > What you suggest is different: light goes through the light absorbant > > material, the fraction that remains after absorption and scattering > bounces > > (reflects) on the white surface of the scanner lid, then goes back > through the > > substance, it is absorbed and scattered for a second time, then it is > captured > > by the sensor. This is not the same thing as scanning a flat, opaque > document. > > > > Sure you will measure something, it won't be in optical density units as > > expected. The standard way of measuring this is using the ratio of the > light > > passing through the material and the incident light, not the amount of > > reflected light. > > > > Some scanners have a light box attachment to scan 35mm slides and film. > You > > probably need one of those to capture bright field images of your gels. > That > > being said, I do not know how accurate the sensors are in office > scanners. > > In addition many scanners have a very short depth of focus, which would > not be > > ideal when scanning a gel that has a finite thickness. > > > > Hope this helps > > Gabriel > > Visiting Professor David Knecht > Beatson Institute for Cancer Research > University of Glasgow > Switchback Road, Bearsden > Glasgow Scotland G61 1BD > UK > -- Joel B. Sheffield, Ph.D Department of Biology Temple University Philadelphia, PA 19122 Voice: 215 204 8839 e-mail: [hidden email] URL: http://astro.temple.edu/~jbs |
In reply to this post by Christian Beringer
Using an internal standard (on the same gel) for normalization should
address these concerns. Ferez >>That >> being said, I do not know how accurate the sensors are in office >> scanners. >> In addition many scanners have a very short depth of focus, which would >> not be >> ideal when scanning a gel that has a finite thickness. > > According to some internet research the scanner sensor seems to be the > same which are used for photometry ! > The resolution is just limited by the CPU frequency and the amount of RAM > of the used computer. > I am not measuring electrophoretic gels, I want to use the > Gel-Analysis-function for quantification of TLC. > So the thickness doesn´t matter, because I am using HPTLC-plates. The > silicagel-layer of these plates is as thin as a a sheet of paper and it is > based on an aluminium-foil. My only problem is the scattering of the > silica-gel. > >>Christian, but why do you acquire RGBs, are you analyzing each channel >> separately > > I think ImageJ allows also to analyse each channel separatly, but until > now I just used the RGB-pictures (.tiff) and the Gel-Analysis-function. I > have no idea how these function works with RGB-scans, but as I tested it > with some .tiff made by Gimp it seems that the colour doesn´t matter for > these function. But tomorrow I will use 8-bit pictures to measure the grey > value. > > The crazy thing is, that the normal TLC-plates (thicker silica particels) > work even better than the HPTLC-plates, I think that belongs to the > scattering, which is increased by smaller particels, which are used for > HPTLC-plates. > Ferez S. Nallaseth, Ph.D. Visiting Scientist Center for Advanced Biotechnology and Medicine Rutgers, The State University of New Jersey 679 Hoes Lane West Piscataway, NJ 08854 Tel: 646 283 5163 (M) Skype Address: fereznallaseth Website:http://sites.google.com/site/nallasethfs |
One more point. Since you are using TLC, and looking at spots rather than
bands, you might want to use the "Analyze Particles" route rather than the Gel Scanning route. By setting a threshold, you can measure the full area and integrated density even if you have variable spot sizes. Joel On Sat, Apr 28, 2012 at 6:06 PM, Ferez S. Nallaseth < [hidden email]> wrote: > Using an internal standard (on the same gel) for normalization should > address these concerns. > > Ferez > > >>That > >> being said, I do not know how accurate the sensors are in office > >> scanners. > >> In addition many scanners have a very short depth of focus, which would > >> not be > >> ideal when scanning a gel that has a finite thickness. > > > > According to some internet research the scanner sensor seems to be the > > same which are used for photometry ! > > The resolution is just limited by the CPU frequency and the amount of RAM > > of the used computer. > > I am not measuring electrophoretic gels, I want to use the > > Gel-Analysis-function for quantification of TLC. > > So the thickness doesn´t matter, because I am using HPTLC-plates. The > > silicagel-layer of these plates is as thin as a a sheet of paper and it > is > > based on an aluminium-foil. My only problem is the scattering of the > > silica-gel. > > > >>Christian, but why do you acquire RGBs, are you analyzing each channel > >> separately > > > > I think ImageJ allows also to analyse each channel separatly, but until > > now I just used the RGB-pictures (.tiff) and the Gel-Analysis-function. I > > have no idea how these function works with RGB-scans, but as I tested it > > with some .tiff made by Gimp it seems that the colour doesn´t matter for > > these function. But tomorrow I will use 8-bit pictures to measure the > grey > > value. > > > > The crazy thing is, that the normal TLC-plates (thicker silica particels) > > work even better than the HPTLC-plates, I think that belongs to the > > scattering, which is increased by smaller particels, which are used for > > HPTLC-plates. > > > > > Ferez S. Nallaseth, Ph.D. > > Visiting Scientist > > Center for Advanced Biotechnology and Medicine > Rutgers, The State University of New Jersey > 679 Hoes Lane West > Piscataway, NJ 08854 > > Tel: 646 283 5163 (M) > Skype Address: fereznallaseth > Website:http://sites.google.com/site/nallasethfs > -- Joel B. Sheffield, Ph.D Department of Biology Temple University Philadelphia, PA 19122 Voice: 215 204 8839 e-mail: [hidden email] URL: http://astro.temple.edu/~jbs |
In reply to this post by Christian Beringer
Christian,
Can you give more details on the scanner? Which resolution do you actually need for the analysis? For controlling the stability of the scanner and calibrate in some cases you can also use IT8 targets (silverfast), both in reflective and transparency mode. I am curious about the images you are analyzing. Can you provide an example image? Send to: [hidden email]. Gary. http://www.gcsca.net Sent from my iPhone On 28. apr. 2012, at 21:41, Christian Beringer <[hidden email]> wrote: >> That >> being said, I do not know how accurate the sensors are in office scanners. >> In addition many scanners have a very short depth of focus, which would not be >> ideal when scanning a gel that has a finite thickness. > > According to some internet research the scanner sensor seems to be the same which are used for photometry ! > The resolution is just limited by the CPU frequency and the amount of RAM of the used computer. > I am not measuring electrophoretic gels, I want to use the Gel-Analysis-function for quantification of TLC. > So the thickness doesn´t matter, because I am using HPTLC-plates. The silicagel-layer of these plates is as thin as a a sheet of paper and it is based on an aluminium-foil. My only problem is the scattering of the silica-gel. > >> Christian, but why do you acquire RGBs, are you analyzing each channel separately > > I think ImageJ allows also to analyse each channel separatly, but until now I just used the RGB-pictures (.tiff) and the Gel-Analysis-function. I have no idea how these function works with RGB-scans, but as I tested it with some .tiff made by Gimp it seems that the colour doesn´t matter for these function. But tomorrow I will use 8-bit pictures to measure the grey value. > > The crazy thing is, that the normal TLC-plates (thicker silica particels) work even better than the HPTLC-plates, I think that belongs to the scattering, which is increased by smaller particels, which are used for HPTLC-plates. |
Free forum by Nabble | Edit this page |