Hi all,
I am currently working with VMTK for level set segmentation on ultrasound data and have encountered several issues and questions regarding parameter adjustments and their impact. I implemented my workflow in Python and used VTI format files for my data. I would appreciate any insights or clarifications from the community. Here are the details:
1. Minimal Differences with Parameter Changes:
- When extracting segmentations for different parameter settings after evolution, there were no significant differences in the results.
- I adjusted parameters such as iterations, propagation, curvature, and advection, but observed minimal differences in the segmentation outcomes.
- When using the test data and pytest provided with the package, there are differences in segmentation outcomes.
Question: Why do parameter adjustments seem to have little effect on my final segmentation results, despite following the same procedures that yielded varying results with the test data?
- For example, when using geodesic level set type and extreme derivative sigma values for the gradient feature image, relative to my voxel resolution (sigma value 1.0 compared to a voxel resolution of 1e-4), the feature image showed minimal gradients. Despite this, the segmentation results remained consistent (as when using a sigma of 1e-4).
I am puzzled as to how the segmentation process can be effective under these conditions, given that it uses the feature image during evolution.
Question: How does the segmentation process remain effective with large sigma values relative to voxel resolution, given the minimal gradients in the feature image?
2. Different level set types:
I experimented with different types of level sets (e.g., Laplacian versus geodesic gradient) without observing differences in segmentation outcomes. I couldn’t find documentation detailing these different types of level sets.
Question: What are the differences between the various types of level sets (e.g., Laplacian versus geodesic gradient), and where can I find documentation on these?
Any insights or suggestions on these issues would be greatly appreciated. Thank you!