Using ImageJ for use-wear studies on stone tools

Please help me!!!

I’m writing my thesis on how stone tools accrue patterns of use (scarring, fractures) after performing particular activity (cutting, scraping) against a particular material (bone, flesh, plant material). I’ve performed 32 activities with different variables.

I’m brand new (5 hours into youtube tutorials) with ImageJ (times also running short), and I was hoping to get some advice on the best way to present this.
Overlapping before and after pictures? Is there a way to do an edge profile instead of an entire surface profile- how can I isolate the edge? My before and after pics (just through a Nikon camera) are slightly different due to marginal lighting differences and thats messed up the surface profiles because the depth appears different when it’s really not.

After use, on a macro level, the edge is worn down, how can I plot that on the software. I’m looking at the use-patterns through SEM as well as 200x light microscopy (but don’t have before pictures, because especially with SEM it would cost me a billion). Can I stack and then compare patterns. Any statistics tests I should run?

I swear I’m not being lazy by asking, it’s just that its 1am and I’m deeply stressing.

I will be so grateful for any general comments, suggestions, or in-depth answers. THANK YOU.

Hello and welcome to the forum!

Could you supply us examples of your data? That would help us understanding your problem.
Just use the upload button in the interface where you write a message. It specifies also which file formats are accepted.

First of all good representative visualizations are the basis for your thesis. So I would start building that first. From that I would define the measurements you want to extract. Such measurements can be done also manually. As long as you have a clearly defined Region of Interest and measurement method.

But that depends on your object of interest. So we can try with some examples :slight_smile:

17%20am 56%20am

Thank you for helping!

So this is the simplest example. Here are before and after pictures of the stone tool. The tool with more shadowing is after, and you can see the edge on the right is more worn than on the before picture. I’ve stacked them, greyscaled them and wanted to see (or plot) how much has been worn. I have closer up images (from a profile view) of the edges, since thats of interest. The rest of the tool doesnt matter. I also edited them on photoshop so the edges probably arnt true to life. Is it better to upload the raw images onto this software (which i’ve included and are those with a darker background- one with scale is before)

I could also try overlap the edges: before and after

I think that is doable. There is a visible difference.

If I were you I would proceed with an initial qualitative assessment.
As a first step crop the images you want to compare and focus only on the rough areas you are interested in. Highlight any strong differences by making smaller crops and blowing them up.
You can show them as insets where you highlighted from which particular area you cropped.
Also use Arrows to point out particular difference.

The biggest difference I can gather from your images so far is the variation in the line.
Here it would be best for the Image Analysis in Fiji that you leave the black background. Use the raw data and not the data that you cropped by Photoshop.

I will see if one can find a good way to segment the line. But I am confident due to the difference in contrast that one can find a good way. I guess some sort of edge detection then segmentation.

The next step is to get a rough estimate of the difference of the lines. Comparing lines with each other is super straight forward. As an initial value I would suggest the standard deviation. Might be sufficient already to visualize a quantitative difference. One can built then on that…

Will get back to you when I have something useful.

One more thing, for image analysis it is in general best to avoid .jpg. Since this format comes with lossy compression. Which changes intensity values and creates spatial artifacts: https://imagej.net/Principles#Why_.28lossy.29_JPEGs_should_not_be_used_in_imaging

Perfect. Yeah comparing the edges by treating them as lines would be great. I’m just not sure how to do that on imageJ as the marking tool doesn’t do much. But I’ll play around with image analysis in Fiji.

Thanks!

Almost there :wink:
I used Fiji: https://fiji.sc

  1. Duplicate before you do any processing:

Image > Duplicate…

  1. Select ROI and focus only on area of interest
    This is important. Select always a comparable area.

Draw ROI and crop: Image > Crop

  1. Detect the edge

Process > Find Edges…

Here it shows that your image acquisition is a bit problematic. You have only a small area of the edge that is properly in focus. Also it is crucial to image the same consistent angle of the edge of your object. That is not to say that the analysis is not possible, it simply increases the error of the analysis. Your results might be small and the variability due to any measurement error might make it hard to see differences.

  1. Get line with Ridge Detection

I used the ridge detection plugin: https://imagej.net/Ridge_Detection.
It is part of the BioRegMed update site. Plugins > Ridge Detection
The Line Width setting was the most important setting with your images. Also remove smaller lines with a minimum line length. Just play around.

This allows you to extract the coordinates of the line. Which one can do all the necessary data analysis.

Amazing. I’ll give that a go and let you know how I go!