I think that it is important to learn about medical image segmentation and classification; many algorithms have been developped in this area. Mazda software is one of the interesting tools.
As said before this is a question that is irrespective of MaZda or any other software. The issue is the how sensitive are your results (in this case using MaZda to calculate a certain metric from the co-occurrence matrix, MaZda is a short of co-occurrence in Polish by the way!) to a certain parameter, in this case size of the neigbourhood. Therefore what you should do is to do a robustness or sensitivity analysis in which you start with a very small neighbourhood and calculate results, then grow it and calculate again and so on. Ideally you should find that there is a sweet spot at which the metrics are at a highest value and then decrease either side. That is the ideal size of the ROI. Of course this can change if you have different resolution or view or other conditions.
Hope this helps.
You can read section 2.2.1 of the following book chapter "Texture Analysis in Magnetic Resonance Imaging: Review and Considerations for Future Applications". It will give you an idea of how to choose the ROI size for texture analysis.
Regards,
Andrés
http://www.intechopen.com/books/assessment-of-cellular-and-organ-function-and-dysfunction-using-direct-and-derived-mri-methodologies/texture-analysis-in-magnetic-resonance-imaging-review-and-considerations-for-future-applications
That is a good book Andres! I wish I could have seen that when I was doing my PhD ;-) In case you are interested in Volumetric Texture, this is a summary of techniques
https://www2.warwick.ac.uk/fac/sci/dcs/people/abhir_bhalerao/publications/volumetric_texture_in_biomedical_image_analysis.pdf
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs