Online Manual for the
WCIF-ImageJ collection

ImageJ
Image Processing and Analysis in Java

8 Colour Image processing

Images can have colour in two ways: pseudocolour or RGB. A pseudo-coloured image is a single channel, (i.e. grey) image that has colour ascribed to it via a “look up table” or LUT (a.k.a. palette, colour table). This is literally a table of grey values (zero to 256 or 4095 whether 8-bit or 12-bit grey) with accompanying red, green and blue values. So instead of displaying a grey, the image displays a pixel with a defined amount of each colour. Differences in colour in the pseudo-coloured image reflect differences in intensity of the object rather than differences in colour of the specimen that has been imaged. For pseudocolour functions see later.

The colours in RGB images (24-bit, 8-bits for each of the red, green and blue channels) and used to show multi-channel images. The colours are designed to reflect genuine colours, i.e. the green in an RGB image reflects green colour in the specimen, the differences in intensity of the green reflects differences in intensity of green in the specimen. There are several RGB functions in ImageJ. Native functions can be found in “Image/Color” and additional RGB functions can be found in “Plugins/Colour functions”.

Fluorescence and transmitted light brightfield images can be merged with the “Plugins/Colour functions/RGB-Grey Merge”  plugin.

Another option would be to use Magenta rather than red in red-green-blue merge. An RGB merged image can be converted to "MGB" with the menu command "Plugins/Colour Functions/RGB to MGB".

8.1 Merging multi-channel images

Currently, only 24-bit colour images are supported, i.e. 8-bit red, green and blue channels. 48-bit colour image (i.e. 16-bit red, green and blue) processing is under development, but currently 16-bit grey images need to be converted to 8-bits before merging. This can be done by adjusting the Brightness&Contrast to the desired level and then using the menu command “Image/Type/8-bit”. Using the “Plugins/LUT/HiLo indicator” LUT can facilitate setting the brightness and contrast: zero is displayed as blue; saturated-white (255) as red.

The brightness and contrast of each channel is best adjusted prior to merging, however, each channel can be individually adjusted after merging with the “Image/Adjust/Color” function. This will not open the control panel if the “Brightness & contrast” control is already open.

8.1.1  Zeiss LSM multi-channel experiments

Each channel in a Zeiss LSM file is imported as separate stacks. For single slice/time point, triple channel experiments this will produce 3 images – one for each of red, green and blue. Dual channel z- or t- series are imported as two separate stacks. To merge the red, green and blue components select Image/Color/RGB merge and put the appropriate image in the appropriate channel.

8.1.2  Biorad multi-channel experiments

Dual channel, 3-D (i.e. xyz or xyt) experiments are saved as two separate files, one for each channel. Each file needs to be imported and then the stacks merged with the menu item: Image/Color/RGB merge. Typically, the red channel should be ***01.pic and the green ***02.pic.

Dual-channel, single frame experiments are saved as single PIC files with two slices. These can be RGB merged with ImageJ menu: “Image/Color/convert stack to RGB”. ImageJ assumes that if it’s a two slice stack, slice 1 is red and two is green and there is no blue channel.

Biorad dual-channel images acquired with CoMOS software have both channels displayed side by side in a single slice. The plugin “Plugins/Colour Functions/Biorad Aligner” will allow these images to be merged. The user can designate the colour for each half of the image from the drop-down option box. Brightness and crontrast of each half of the image can be adjusted separately by selecting each half of the image, adjusting and then applying the Brightness and Contrast change. To facilitate this, the keyboard function key “F4” selects the left hand image panel when pressed once, then the right hand image panel the next time it is pressed.

8.1.3   Interleaved multi-channel experiments

Multi-channel experiments acquired on Noran and Perkin-Elmer systems are imported with the different channels interleaved, i.e. slice 1 is timepoint1-channel1, slice2 is timepoint1-channel2. The stack needs to be “De-interleaved” before it can be RGB-merged. This can be done with “Plugins/Stacks-Shuffling/DeInterleave” and entering the number of channels in the dialog (typically “2”). The two stacks can then be merged via: “Image/Color/RGB merge”.

8.1.4  Colour merging

An alternative to the normal Red-Green merge is to merge the images based on Cyan and Magenta, or Cyan-Yellow or any other colour combination. 

This can aid visualisation of colocalisation due to our poor perception of red and green colours. The “Plugins/Colour functions/Colour Merge” function will perform a ‘difference’ arithmetic processing on the image stacks you select. This is not strictly a merge (when cyan and magenta merge they produce white, not yellow) but facilitates visualisation of the separate channels (See Demandolx and Davoust, J. Microscopy, 1997 v185. p21). You can perform a true merge if you turn off the “Difference” option.

Run the plugin and select the two images to be merged. Select the desired colours from the drop-down options. <Current> uses the LUT that the image currently has (this is often the desired LUT). The “Difference” option performs a “difference” arithmetic operation rather than a “addition”. If the “Pre-sub 2 from 1” option is checked the second image is subtracted from the first prior to merging. This may be useful for merging grey brightfield images with fluorescence where simple addition can result in washed-out look t the fluorescence image. It can look “unnatural” though. Another option may be to “darken” the brightfield image prior to merging (use menu command “Process/Math/Multiply” and use the value 0.5-0.75).

8.2 Splitting multi-channel Images

Sometime a merged RGB image may need to be displayed along with the separate channels in a final figure

.

The plugin "Plugins/Colour Function/RGB to Montage" works with single slice RGB images. A new RGB stack is created, channel1 being the red channel; 2 the green and 3 the blue. The fourth slice is the merged slice. If a channel is empty (e.g. a red-green merged image will have no blue) then the slice is omitted. The then plugin prompts you for the layout of the montage. You have the option to alter the width of the white border between the image panels.
If this dialog is closed then no montage is made and the stack is kept open.

8.3          Pseudocolour

Judicious use of LUTs can be very useful in highlighting the desired features of an image. The human eye can only perceive relatively few shades in one scene. Pseudo-colouring images can make the data more visible

Have a play and see which LUT helps illustrates the features in your image.

Montage compiled from a stack generated using the plugin “Plugins/LUT/List LUTs.

Different LUTs are available via the menu commands “Image/Lookup table” and also “Plugins/LUT”. Custom LUTs can be made with the “Plugins/LUT/LUT panel”. Extra LUTs can be found in C:\ImageJ\LUT and can be applied by highlighting your image and selecting “Plugins/LUT/OpenLUT ” then selecting the required *.lut file. Opening LUTs via this command rather than other ways (i.e. “File/Open…” or “File/Import/LUT…”) prevents the default folder from changing to “C:\ImageJ\LUT”. Custom LUTs can be made and edited using the “Plugins/LUT/LUT panel ” plugin.

When using a non-standard LUT it may be useful to add a greyscale ramp so the user can judge which colour represents which grey-scale value. A ramp can be added with the native function "Analyze/Tools/Calibration Bar"  or by making a selection the size and position that you want the ramp and running the plugin "Plugins/LUT/Add Ramp". If there is no space for teh ramp, try enlarging the image canvas with the menu command "ImageAdjust/Canvas Size".