Radiomics, contrast and difference variance

The code in imagej-ops is very useful and helpful. I do have a problem however.
The code for contrast and difference variance looks exactly the same to me.
I can recognize the contrast from the Haralick paper, but the difference variance is not as obvious.
I think there may be a bug in the code.

I would be happy to be wrong, but so far it looks the same.

Ilan

Hi @ilan

Thanks for pointing that out! DefaultContrast is not what it should be. I will provide a fix.

Best

1 Like

Thanks for the great software and the very fast reply.
To me DefaultContrast looks OK.
DefaultDifferenceVariance looks questionable.

Ilan

Based on this: http://murphylab.web.cmu.edu/publications/boland/boland_node26.html DefaultDifferencevariance looks OK. I am double-checking with this paper: http://haralick.org/journals/TexturalFeatures.pdf

I changed the implementation to represent the Difference Variance from the paper: http://haralick.org/journals/TexturalFeatures.pdf

PR is out :slight_smile:

1 Like