From left to right: original image, 8-bit greyscale, threshold image, particle analysis.
Last Fall I worked with Glor Lab graduate student Julienne Ng to develop a method of measuring the number and size of scales on Anolis lizards. We are hoping to determine if a relationship exists between temperature and humidity of a habitat and scale size. Below is the method that we have developed using ImageJ, which approximates both the number of scales and the area of those scales.
- Open a .jpg image of the ventral side of a lizard specimen in ImageJ. Make sure that your image includes a scale of known length.
- Draw a line across one unit of your scale.
- Select “Analyze – Set Scale.” Enter the known distance and unit. The measurements on your toolbar will now use your established units.
- Select an area of the lizard stomach that is 0.30 cm x 0.30 cm. Select “Image – Crop.”
- Now that you have a standardized image, you need to change the format of the image so that scales can be clearly distinguished. Select “Image – Type – 8-bit.”
- Select “Image – Adjust – Threshold.” Select “Dark background” and click “Apply.” Your image should now be a black and white pattern of scales.
- Select “Analyze – Analyze Particles.” In the next window, set the particle size to a range of 0.0001 – infinity. Make sure that “Display results” and “Summarize” are checked. Select “Show: outlines.”
- A new window will show a new version of your image, indicating each numbered scale. The Results window will list each scale and its area, and can be exported as an Excel spreadsheet for further analysis.
This process is obviously still in a preliminary stage, but we’re very excited about the progress that we have made so far. Has anyone else ever tried something like this? Does anyone have advice on improving our method?
Latest posts by edunn1205 (see all)
- Using ImageJ to Analyze Scales - June 1, 2012
- Holiday Gift Ideas for Anole Lovers - December 4, 2011
gabriel gartner
Hi,
We have employed a nearly identical method here in the Losos lab for a very similar project so we should chat! I have also simply counted ventral scale rows under a dissecting scope (which is tedious, but very repeatable).
One issue I had with the imageJ technique, though more for dorsal scales than with ventrals is that unless the lighting is very even, you will have problems with thresholding step and isolating individual scales. Have you guys had much success with this?
Cheers,
Gabriel
Julienne Ng
Glad to hear that you also came up with a similar method! So far we have just focused on ventral scales mainly because we are scanning specimens on a scanner bed and the ventral side is the only side we are able to get on a flat plane using this technique. However, we initially did play around with some dorsal scans and certainly also had a harder time with the darker scales and the threshold step. We haven’t invested much time on this though so unfortunately haven’t come up with a solution!
Tamara Fetters
Hi,
I’ve been tinkering around with pictures of dorsal scales and the best way I’ve found to isolate individual scales is to follow the directions above until you complete step 4 (Select an area of the lizard stomach that is 0.30 cm x 0.30 cm. Select “Image – Crop.”) At this point, I would select “Process – Sharpen”. Then continue as above until you complete step 6 ( “Select “Image – Adjust – Threshold.” Select “Dark background” and click “Apply.”). After this step, I would select “Process- Binary- Watershed.” This should separate particles that have merged together by adding a line where it feels separation should occur. Then, when you continue to analyze the particles, scales should be more clearly defined. This also allows you to not worry so much about completely isolating the scales during the threshold step, so you may be able to get more accurate scale areas.
I know this thread is from over a year ago, but I stumbled upon it while searching for a way to measure the number/area of scales and thought this information might be helpful to others.
Cheers!
Tamara Fetters
Also, keep in mind that whether or not you want to sharpen the image greatly depends on the image you have. Sometimes it’s best to sharpen it, sometimes it’s best to adjust the contrast, and sometimes it’s best to just leave it alone.