ID Secordary not grabbing full ROI

Hi Guys,

I'm trying to use CP to recognize a myelin sheath in a TEM cross-section image of axon bundles. I'm doing some voodoo to simulate a fluorescent image and for the most part it works. The problem I'm running into is that when recognizing the actual myelin, the IDSecondary module isn't expanding enough to grab the entire region, but it should. I can't for the life of me figure out why it's segmenting where it is. The attached image shows the result no matter the settings I use in ID Secondary.

Any suggestions?

-J


Hi,
Could you post the raw image that goes along with this result, plus the pipeline that you’re currently using?
Thanks!
-Mark

Here is a sample. The goal is to get the inner diameter and outer diameter of the bolded rings. I’m very curious to see if you can improve it!
AxonPerimeterPipeline.cp (10 KB)

Sorry for the bump, any suggestions on how to expand?

-J

Hi J,

I’m attaching a pipeline that might get you a bit closer to what you want. The difficulty is that IdentifySecondary typically relies on some amount of smooth variation in image intensity for the propagation to work well; the myelin is fairly uniform in intensity, so any small amount of variation caused the propagation to halt prematurely.

I tried to get around this by taking the distance transform of the binary thresholded image; the result of this is the introduction of a smooth variation as a bright ridge at the halfway point in the myelin region to use as input for IdentifySecondary. Then, I expand the secondary object manually to try to get the rest; unfortunately, I’m at a loss as to how to do this automatically.

Regards,
-Mark
2012_10_11.cp (8.38 KB)

Hey Mark,

Thanks for the pipeline. Of course it’s ideal if it propagates to the edge by itself instead of any finessing. I think that if it is able to more accurately find the middle of the myelin than I can find the difference of the ID and OD then just double it. That may be a more accurate and automatic way. So this pipeline could be very useful.

Thanks for your assistance on this!

-Joshua