A general remark: Fluorescent biological samples are obviously coloured (unlike unstained samples observed with transmission illumination). Also, colour images are often more appealing than black-and-white images. However, digital monochrome images do not only require less storage space (see
Bit Depth of Digital Images): it is also simple to convert a black-and-white image into, say, a black-and-green image with any common imaging software (see
Displaying Digital Images). Consequently there is no real need to acquire colour images if the sample is labelled with only one fluorophore. For multi-labelled samples there are different ways to get colour images:
Sequential Multi-Pass Imaging With Monochrome Cameras
At first sight the possibility to generate colour images with black-and-white cameras might sound strange but this is probably the most common way in fluorescence microscopy. Let's consider a sample triple-labelled with DAPI, FITC and Texas Red. If the microscope is equipped with the corresponding three different filter cubes or, better, a filter wheel or monochromator to switch the excitation wavelength and a matching triple-band filter set, then the red, green and blue fluorescence can be excited subsequently and three different monochrome images acquired. If the three data sets are fed into the three colour channels (red, green and blue) of a 24 or 48 bit RGB image, the result looks just like a one-shot colour image as the following example illustrates.
The distribution of the different data sets into the three channels is not that simple if the emission does not match one of the three primary colours (RGB). Other colours can be generated by using the principles explained in page Colour Images. For example, by displaying a sub image equally bright in green and blue, it will appear cyan.
Sequential Three-Pass Colour CCD Imaging
In this type of imaging system the (colour) camera features either a rotating filter wheel or a rapidly switchable liquid crystal array screen to capture the red, green and blue colours subsequently. The latter type of cameras has the advantage that the colour switching is faster and vibration free. The three individual images are recombined off-chip just like in the method above.
Three-Chip CCD Cameras
These cameras employ an optical train featuring a beam-splitting prism and trim filters to split the captured light into three colour components and project each onto a different CCD chip. The three colours are then recombined off-chip as in the sequential colour imaging techniques. The fundamental difference here is that all colours are imaged simultaneously.
Matrix Filter Colour CCD Imaging
A camera with a matrix filter system features either a CCD or a CMOS chip where the photodiodes are covered with a regular pattern of red, green and blue colour separation filters (RGB Bayer mosaic filter). Repeating subunits consist of four photodiodes arranged in a square and masked with one red, two green and one blue filter, corresponding to one full colour set. Two green filters are used because green is at the centre of the visible wavelength region and the human visible response reaches its maximum here. In higher resolution cameras the second green filter in the matrix is often replaced by a teal (blue-green) filter. With digital image processing, a colour image can be assembled like in the methods described above. The caveat of the matrix filter is the rather low transmission and the resulting low light sensitivity of the cameras. As a consequence they are of limited use for fluorescence microscopy. CMY filter arrays are also being used sometimes for improved spectral response.
Because each photodiode contains the information of one of the three colour channels only, interpolation algorithms have to be used for the colour reconstruction of each pixel. If the data in the four photodiodes that form a red-green-green-blue subunit would simply be combined, a loss of spatial resolution similar to 2x2 binning would result. Two out of a variety of techniques are shown below. The left scheme depicts how the average counts of neighbouring pixels of different colour are used to add the two missing channels to each pixel for optimised spatial resolution. A somewhat more complicated algorithm is used for optimised colour resolution as shown in the right drawing. Here virtual pixels are assembled each containing the average of two adjacent green pixels and parts of the counts of 4 red and 4 blue pixels weighted by the distance. One set of blue or red arrows is not shown in each case for clarity.
The distribution of the different data sets into the three channels is not that simple if the emission does not match one of the three primary colours (RGB). Other colours can be generated by using the principles explained in page Colour Images. For example, by displaying a sub image equally bright in green and blue, it will appear cyan.
Sequential Three-Pass Colour CCD Imaging
In this type of imaging system the (colour) camera features either a rotating filter wheel or a rapidly switchable liquid crystal array screen to capture the red, green and blue colours subsequently. The latter type of cameras has the advantage that the colour switching is faster and vibration free. The three individual images are recombined off-chip just like in the method above.
Three-Chip CCD Cameras
These cameras employ an optical train featuring a beam-splitting prism and trim filters to split the captured light into three colour components and project each onto a different CCD chip. The three colours are then recombined off-chip as in the sequential colour imaging techniques. The fundamental difference here is that all colours are imaged simultaneously.
Matrix Filter Colour CCD Imaging
A camera with a matrix filter system features either a CCD or a CMOS chip where the photodiodes are covered with a regular pattern of red, green and blue colour separation filters (RGB Bayer mosaic filter). Repeating subunits consist of four photodiodes arranged in a square and masked with one red, two green and one blue filter, corresponding to one full colour set. Two green filters are used because green is at the centre of the visible wavelength region and the human visible response reaches its maximum here. In higher resolution cameras the second green filter in the matrix is often replaced by a teal (blue-green) filter. With digital image processing, a colour image can be assembled like in the methods described above. The caveat of the matrix filter is the rather low transmission and the resulting low light sensitivity of the cameras. As a consequence they are of limited use for fluorescence microscopy. CMY filter arrays are also being used sometimes for improved spectral response.
Because each photodiode contains the information of one of the three colour channels only, interpolation algorithms have to be used for the colour reconstruction of each pixel. If the data in the four photodiodes that form a red-green-green-blue subunit would simply be combined, a loss of spatial resolution similar to 2x2 binning would result. Two out of a variety of techniques are shown below. The left scheme depicts how the average counts of neighbouring pixels of different colour are used to add the two missing channels to each pixel for optimised spatial resolution. A somewhat more complicated algorithm is used for optimised colour resolution as shown in the right drawing. Here virtual pixels are assembled each containing the average of two adjacent green pixels and parts of the counts of 4 red and 4 blue pixels weighted by the distance. One set of blue or red arrows is not shown in each case for clarity.
Monochromatic colour information of neighbouring red, green and blue pixels on a matrix filter CCD chip is averaged using different algorithms to generate the coloured pixels of a RGB image.
