I have a grayscale image of which the original RGB image was formed by combining Green and Red channels which represent different types of cells (live/dead). Splitting the channels in RGB mode is giving Red, Blue and Green to all the cells. I want to confine green to the live cells and red to the dead cells using ImageJ.
Hi Krishnaraj,
I'm confused regarding your live/dead assay. Wasn't it performed using fluorescence? If so, the images (red and green) were not RGB but monochrome (grey scale)
your original image (from whatever microscopy your were using) should be a combination of two independent monochrome channels. if you save the data in the right format (not merging anything at priori) you should be able to use fiji to split the channels. If you save your data as a 'merged composite' (which is apparently what you're doing) then you miss the information about each independent channel. Not really sure if I'm answering the question, but at least this is what I understood was the problem
If you have a single channel grayscale image there's probably nothing you can do to get the original color channels. Such image contains only the information about intensity averaged across all three channels per each pixel.
If you change the format of your grayscale image and treat it as an RGB, splitting will give you the same image for red, green and blue, all identical to the original grayscale image. This is because only a mix of these 3 colors in equal and specific intensities results in the same corresponding intensity of white = certain shade of gray in every pixel ("grayscale" image).
If you had an RGB image made by merging of original red and green images, each pixel would still contain information about the intensities of red, green and blue channels separately. You would get the original channels by splitting the RGB image with one difference: If the original data were e.g. 12- or 16-bit, splitting an RGB image will provide only 8-bit result (ImageJ doesn't work with higher quality RGB in default, there might be some plug-ins), which means that you are losing information and quality. You could discard the blue channel which would be all black = zero intensity values in your case. I just tried all of this with one of my images and it functions the way described. Note that this method can only be used if the original image was made of red, green and blue channels or their subset, nothing else!
Anyway, this should be the last emergency solution to be used only when
a) you work with an image from someone else (watch for copyright) and you don't have an access to the original data or
b) you've lost your own original data.
Of course, the best practice is to keep your original data before any processing (and possibly some intermediate processing steps) ideally with a back-up and use those.
- Badges
- Science method