Hi Folks,
I'm doing a very simple assay which involves placing plant cuttings in a red dye solution. The dye can then be seen in the leaf (due to transpiration).
I'd like to try and make this more quantitative by taking photos of the leaves and then comparing the control (no dye) to the experimental (with dye).
I've been playing around with imageJ using the various functions, but I can't seem to find anything that works well. I think one of the problems, is the background 'green' color of the leaves, which varies in intensities from leaf to leaf. In other words:
red dye + light green leaf = red is very 'red' and thus easily picked up by software as being 'red'.
red dye + dark green leaf = red is a darker 'red' and not as readily picked up as being the color 'red' by the software.
red dye + really dark green leaf = red is almost a black color and is not picked up by the software. However, visually (qualitatively), you can still see the that the leaf is indeed red.
I guess the easy way around this is to initially pre-select leaves that have the same level of green (intensity). However, this is kind of cumbersome, and I don't want to spend a lot of time selecting plants.
Any thoughts?
Thanks a bunch!
Christian
Hi Christian!
Do you already know about WIMASIS? We have already developed custom solutions for staining quantifications, like our Hairy Roots Solution, which analyzes the root growth as well as the local concentrations of the metabolites:
(https://tu-dresden.de/ing/maschinenwesen/int/ressourcen/dateien/bioverfahrenstechnik/bt_projekte/IVK_WBTWPC/IVK_Lenk_02_PictRecogn_en.pdf?lang=de)
If you want to give us the opportunity to help you out, just initiate a free Feasibility Study in our myWim Platform (https://mywim.wimasis.com/)
How exactly are you doing the analysis in ImageJ? Which plugin/function are you using? Are you using scripts?
Have you tried to split the RGB channels, then analyze only the red channel? (Go to Type--> RGB stack to split a RGB image from a camera into 3 channels). That might help with your color contamination problem if you are simply looking at total red area...
Another option is to convert your RGB image to HSB image (Type --> HSB stack). The Hue channel (first channel) gives a code to represent the color, independent of it's color intensity. See https://en.wikipedia.org/wiki/HSL_and_HSV .
If you are looking at red intensity, however, it might be trickier since you cannot disentangle the background color from the red dye, short of introducing a red filter to your camera or microscope... Some red dyes are also fluorescent (an accurate quantification with a confocal would work in this case). Evans blue dye is a fluorescent dark blue dye that is also easily seen by eye, but fluoresces red in the presence of proteins.
Another solution is to use photoshop's white balance functions (Image--> Adjustments-->Levels)... though it is not recommended for beginners...
Doing quantification with a dye or color is very tough job. The reason spectrophotometers can do quantification is that they measure absorption of certain wavelength and not the color itself. I can not really recommend you to use red color for quantification.
Of course you can just split RGB channels (can do with even photoshop) and then find the gray scale value of the Red channel only. But this will not give you the true estimate as the perception of color varies with factor other than the wavelength of light.
A better solution is to use a fluorescent dye. and then measure the fluorescence. This will give you values that are comparable and so good for statistics such as a t-test.
Regards
Have you tried to split the RGB channels, then just use the green channel as reference? Try to substract the green image from the red one.
- Badges
- Science topic
- Similar topics
- Engineering
- Chemical Engineering