User Tools

Site Tools


Inka's data

A basic Python course for biologists

Inka's analysis

Continued on Luis development by scanning the segmentation parameter space. It looks like their data gets better results with a much larger width on the SG-filter (9+). I got comparable segmentation results to our old confocal data with the same strain (ca 90-110 neurons). I changed thesignal extraction to not have a filter and built it into the volume class. Now extraction is very fast.

For comparison with Luis' results. For some reason the 21z Plane version performs poorly. I expect this is the same animal and thus it was prebleached.


  • Need to run a segmentation benchmark on ground truth/human annotated data
  • Explore parameter space
  • parameters should have real-world relation to image resolution etc.
wiki/documentation/monika/python_course.txt · Last modified: 2019/11/05 00:24 (external edit)