Dear Shane.
Your eyes have 3 color sensors, red green and blue,
your computer screen had red green and blue pixels
(and in printing its cyan, yellow, magenta and black)
So, now if you want to merge 3 colours,
thats fine, make each colour channel one of red green and blue,
It will look very pretty (hopefully).
Now with 4 colours you are faced with a big problem.
Your eyes are not capable of picking out 4 different colours
out of an RGB image.
As far as your eyes are concerned,
all colours are made of some mixture of red, green and blue.
So if you already have a red, green and blue channel,
then try to add 1 more, what colour would you pick?
Magenta... no, because thats a mix of red and blue.
Similar for cyan and yellow, and other colours.
So, if you do mix red green blue and magenta,
you will end up with a pretty image,
but it will be impossible to tell which parts
of the red part of the image come from
the red channel or from the magenta channel.
Some people use
red, green blue and white,
but you are faced with that same problem .
There cold be a place with almost equal red green and blue intensity
(= white)
which you could not tell from a white pixel
If you are just trying to make a pretty picture,
and not say anything quantitative or too specifically qualitative,
you might get away with RGB+white.
If i reviewed it or saw it in talk,
I would raise the points made above.
If you are trying to look for colocalisation
(or more specifically spatial intensity correlation)
you can do this in pairwise way,
using the imageJ colocalisation plugins
(read the docs at
http://www.macbiophotonics.ca/imagej/
)
Looking at a 2D intensity histogram / scatterplot
of once channel vs a second gives a much better way
of visualising the spatial intensity correlation between then
than a colour merge image.
In my opinion, colour merge images are hard to interpret,
can be misleading, are a poor way of visualising "colocalisation"
and this researcher would send back any paper to an author
who has used colour merge images to try to show
between channel spatial intensity correlations (colocalisation)
In fact I would go as far as to say,
wrong interpretation of colour merge images could
ruin you career.
2D hisotgrams / scatter plots are a much better way to show it.
cheers
Dan
Date: Sat, 10 Nov 2007 16:25:55 -0500
Subject: 4-colour merge?
Hi All,
I'm a relatively new ImageJ user and would like to merge 4-colours
into one composite image (CY3, CY5, Alexa488 and DAPI). Is this
possible in ImageJ? How would I do this?
Thank you,
Shane
Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Processing and Analysis
Light Microscopy Facility
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany
+49 (0)176 61262128 (German Mobile)
+49 (0)351 210 2627 (Work phone at MPI-CBG)
+49 (0)351 3265048 (German Festnetz "landline" - cheaper)
+49 (0)351 210 1078 (Fax MPI-CBG LMF)
+358 (0) 468102840 (Finnish mobile, only when I'm in Finland)
http://www.bioimagexd.net
http://www.chalkie.org.uk
dan-***@public.gmane.org
( white-***@public.gmane.org )