Revision as of 16:04, 6 September 2013

This utility will analyze photographs of Ketostix to provide a better estimate of the level of ketones in the urine than is possible by eye alone. This utility is at an early stage of development and should be considered somewhat experimental. Let me know if you find it useful and I may continue to develop it further. You can download the utility from http://fellrnr.com/Utilities/KetoAnalysis.exe

1 Ketostix or Multistix?

You must use the color chart from the Ketostix; the color chart from other test strips is not supported. However, you can use either the Ketostix or the Multistix version that also checks other parameters. I prefer the multi sticks as they check other important variables, including urine pH, urine specific gravity, and the presence of blood in your urine.

2 Taking the photo

This utility is quite sensitive to the quality and the layout of the photograph being analyzed.

The photo must contain both the Ketostix and the color chart for comparison.
The photo must be well lit, but without any reflections on the color patches.
The background must be white.
The Ketostix placed to the right of the color chart.
The color chart must be arranged with the trace color block on the left and the highest intensity color block on the right.
Only the five positive color blocks should be included on the image, not the negative color block.
The image needs to be quite close up. I have used an iPad and a Dell Latitude 10. The Latitude 10 is ideal as you can run KetoAnalysis on the tablet so you don't have to transfer the image.

An example image is shown below.

This is a sample image, showing the color strips and ketone sticks. Note the number of color strips shown, and the position of the Ketostix.

3 Using the utility

After taking an image of your Ketostix, start the utility and use "File", "Open" to load the image. If the utility is able to analyze your image successfully, you'll see something that looks like this.

The highlighted area contains the results of the analysis. There are three approaches to analyzing the results; the overall average, the pixel at the center, and pixels around the center. The last of these appears to be the best and most reliable.
These radio buttons control which version of the image is shown. You can see the original image, the image that has been filtered (the white background stripped out), and the image that has the target areas highlighted and numbered.
These radio buttons control what the utility does a startup. You can set it to reopen the last file, reopen the latest file in the directory last used, or do nothing.
This utility uses filtering to work out where the color patches are. The six numeric input fields define the scope of the filtering based on hue, saturation, and luminance. You can click the fields directly, or hit one of the two search buttons for the utility to try various values. At the bottom of the highlighted section you will see the details of the pixel under the cursor, which may help find the right filter parameters.