Get Your Brights Brighter & Darks Darker: CLAHE + diceCT

By Aki Watanabe (@akiopteryx)

Contrast limited adaptive histogram equalization (CLAHE) is a procedure for enhancing local contrast in an image or stack of images. In contrast to standard histogram equalization that applies single formula for enhancing contrast across the entire image, CLAHE applies multiple equalizations within partitions of an image, resulting in more localized and subtle contrast enhancements. This results in digital contrast enhancement that is not dominated by overly deep blacks or excessively bright whites.

For diceCT, CLAHE is very useful for improving edge recognition for digitally segmenting regions of interest (ROI) based on your CT data. It is particularly helpful when applied to sub-optimally stained specimens.

CLAHE is implemented in FIJI (ImageJ) and the script is available freely and openly. To perform it on a stack of CT images:

  1. Drag and drop the folder that contains a stack of CT images into FIJI (download here: https://fiji.sc). Wait until FIJI reads in the entire image stack.
  2. Copy the CLAHE script from the “Tips” section on the website.
  3. In FIJI, go to “Plugins” > “New” > “Macro”. A new window will open. In the text field, paste the CLAHE script.
  4. Select “Run”. One can specify the “block size,” “histogram bins,” and “max slope” parameters (the details of which are outlined on ImageJ.net), but I have found that the default parameters do a fine job.
  5. Once CLAHE has gone through the entire stack, save the new image stack in a different folder to keep the original CT image stack intact (you never know when you’ll need them).
  6. In CT data processing program of your choice (e.g., Avizo, VGStudio), read in the modified image stack.

Here is a before and after example to illustrate how CLAHE enhances diceCT images.

 

Screen Shot 2016-08-01 at 6.22.53 PM
DiceCT, two-year old Alligator mississippiensis head in transverse views (left) unmodified, and (right) filtered using CLAHE.

CLAHE processes image stacks fairly quickly, so I recommend trying it with all diceCT image stacks. Generally, it improves edge recognition for all 3-D rendering programs, thus, greatly reducing the time spent on segmenting ROIs. As mentioned above, CLAHE is particularly useful for sub-optimally stained specimens, which is helpful when one cannot devote time or reserve frequent CT scanning sessions for checking and optimizing the stain concentration and duration.

Potential issues with CLAHE include the increase in file size associated with CT data from having both original and modified image stacks. In addition, CLAHE may accentuate unwanted artifacts like beam hardening, so there is further motivation to minimize such scanning artifacts.

Those interested in learning more about the method can visit Wikipedia and read through the original article (Zuiderveld, 1994).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s