Analysis of the area covered and/or the intensity of a given stain can be a powerful tool in image analysis. It can also form the basis for more complex analysis.
Before analysis of an individual stain/channel can occur it may be necessary to separate out the colour channels as individual images. This will only work for an image that was created by merging 3 fluorescent images together. It is not possible on standard colour images (H and E, DAB, etc.) or on merges of more than 3 fluorescent channels (unless it has been saved as a hyperstack/multi plane tiff).
Open Fluoro 01.tif from the Demo Images\Widefield Images\Fluorescence Measurement folder
2. To separate the channels go to Image 🡪 Colour 🡪 Split Channels
3. The image will be split into 3 images. Each image in this case is only an 8bit monochrome image but will be suitable for basic measurement
For an image to be able to be measured it first needs to have a threshold applied to it. Select the red channel image (blood vessels). Go to Image 🡪 Adjust 🡪 Threshold
2. Tick the dark background box and leave all other settings at default. A red mask will be placed on the parts of the image that are selected by the threshold. You can adjust the threshold by moving the sliders. The top slider sets the bottom range of the threshold and the bottom slider adjusts the top range of the threshold. The red box on the histogram shows which parts of the histogram are being thresholded.
If you press the auto button Fiji will attempt to automatically threshold the image. The drop down box on the left (the one that says default) has 16 different auto threshold algorithms to choose from.
Fiji has a module that will allow you to easily test each auto threshold algorithm so you can decide which the best to use for your sample is.
Go to Image 🡪 Adjust 🡪 Auto Threshold
2. In the dialog box that opens up set the method to Try All, make sure the White objects on black background and Show threshold values in log window boxes are ticked. Press OK.
3. The resulting montage shows what each auto threshold algorithm would produce on the image. Each frame of the montage represents a different algorithm which is listed in the log that was created.
Go back to the original red channel image and set a threshold on it. The default one is fine.
2. Go to Analyse 🡪 Measure. A window similar to the one below will open up
By default the measure command does not take any notice of the threshold. The current data measured shows the area of the whole image (since this is an uncalibrated image we can tell it was taken with a 1.92 megapixel camera), the average intensity of the whole image and the minimum and maximum intensity values.
3. Right click on the results window and select Set Measurements.
NOTE: You can also go to Analyze 🡪 Set Measurements…
4. The resulting window allows you to select what measurements you wish to make. It also allows you to limit the measurements to only thresholded areas. Select Limit to Threshold and Area Fraction. Set as below and press OK.
5. Now remeasure the image. The values for area, mean grey value and min and max grey values that are measured this time apply only to the areas under the threshold. The results can be saved to excel or a txt file if required.
NOTE: The value for area is in pixels as this image is uncalibrated.
For brightfield images, like those stained with DAB chromogen or H and E, the area of stain in an image can be measured by thresholding like above. It is a little more complicated as chromogens, especially DAB, are not a pure colour but made up of a mix of red, green and blue.
The easiest way to be able to measure this brown channel is to extract it out into its own image using colour deconvolution. Colour deconvolution allows the separation of colours by an automated threshold method. The result is not a threshold you can measure but 3 separate images representing the “channels” of the original image. These “channel” images can then be treated like single channel fluorescent images.
NOTE This technique is only good for counting objects or measuring areas, it cannot be used to measure intensities of the stain due to the limitations of how enzyme based chromagenic stains work.
Open DAB 01.tif from the Demo Images\Widefield Images\DAB
2. Go to Image 🡪 Colour 🡪 Colour Deconvolution
3. In the resulting window there is a choice of options in the pull down menu. The first option, From ROI, lets you specify the different stains in your image manually. The others are preconfigured to work with most standard histological stains. For this example select H DAB and press OK.
4. The original image will be split into three images.
5. The three images represent the haematoxylin stained nuclei (Blue, Colour_1), the DAB chromogen (Brown, Colour_2) and what is essentially a mathematical remainder but may be useful for some things. This third channel may contain another stain if the original image was a 3 channel stain (e.g Mason’ Trichrome).
6. The brown channel image can now be thresholded and measured as if it was a single fluorescence channel. Make sure that Dark Background is not checked this time though.
Since colour deconvolution essentially separates the chromogen stain from the nuclei signal it is possible to turn the results into something that resembles a fluorescent image that may make it easier for people to see the results of the stain.
The first step is to invert the colours so that dark becomes light like in a fluorescent image. Select the nuclei and brown channels in turn and go to Edit 🡪 Invert
2. Now apply a blue LUT to the blue image and a green LUT to the brown image
3. These images can now be combined into a composite image like shown before
4. The result is a pseudo fluorescent image of the original DAB stain