high resolution images cause jaggedness

[royceyork]

Hello Matheus,

I've been using mitograph to analyze yeast mitochondria. It's been working quite well at low res (0.167um _xy) but once the xy resolution goes below a certain limit, mitograph becomes unreliable. Below is some mitograph output from the mitos of two cells. The first two images are from a low res image, the skeleton looks nice and smooth and the nodes look appropriately placed at mito ends and junctions. The second two images are from a much higher resolution image; now the skeleton output is much more jagged and way too many nodes are assigned. This completely ruins the quantification.

Things I've tried (on the input as I don't know enough C++ to get under the hood of mitograph):
zoom (from scipy)- significantly helped the jaggedness but changes geometry so I'm not a fan
smoothing, gaussian or median filters in imageJ -- no effect

appreciate any advice! let me know if I can provide any more information. Thanks!

[vianamp]

Hi. Thanks for using MitoGraph.

Could you please tell me what is the pixel resolution of the last example?

[royceyork]

-xy 0.0657 -z 0.3 um

[vianamp]

Yes, MitoGraph was design to perform numerical computation on images of tubular structures with diameter of about 3 to 6 pixels across. Because your high res images don't meet this criteria, MitoGraph is creating those spurious nodes and edges. The only thing I can think of is resizing your data, but it seems you are doing that already with the zoom function?! I would downsample the image to about 40% of the original size in XY and give it another shot. You can also try to increase the scale range -scales 5 6 5, but this is unlike to help much. Let me know what you think.