Focal Adhesion

Does anyone have a pipeline for detecting Focal adhesions and their properties like size, number, intensity, length ???



FAD analysisPIPE zyxin.cp (5.17 KB)

Hi,

I would recommend placing an EnhanceOrSuppressFeatures module before he IdentifyPrimaryObjects module. You can use this module to enhance the adhesions by setting the operation to “Enhance”, the feature to “Speckles” and the feature size to approximately the size of the focal adhesion in pixels (or a little larger, in your case, maybe around 20). Afterwards, feed the result into IdentifyPrimaryObjects. You are probably better served changing the thresholding method to RobustBackground or 3-class Otsu global with the middle class set to background and adjust the threshold correction factor to suit.

Hope this helps!
-Mark

Many Many thanks for your suggestion…works beautifully!!! I am very impressed! It would have taken me a lot of time to figure out…thanks to you I am up and running in a short amount of time. I will post my pipeline soon for everyone to use. May be you could put it as example pipelines.
PDU

hi pduchil,

I’m about to begin processing a whole load of data, and detecting the focal adhesions would be a huge bonus. how did you get on in the end using CP to detect them? Any further tips you might have for building my pipeline?

Thanks,
Paul

Hi Paul,

If pduchil hasn’t responded yet, and if you still need assistance, you are welcome to post some sample images and tentative pipeline to this thread.

Regards,
-Mark

Hi Mark,

Thanks for getting back to me, notifications were turned off so I’ve only just come across your reply. I had some success with your suggestions, and I’ll definitely post some images and sample pipeline up here when I get a chance (asap)

Cheers,
Paul

Hi Mark,

Here’s an idea of what I’m getting so far:

As there is quite a bit of noise from the focal adhesion stain (anti vinculin) I’m having some trouble getting the thresholding right.

Starter Image


EnhanceOrSuppressFeatures
Operation: Enhance
Feature Type: Speckles
Feature Size: 15


IdentifyPrimaryObjects
Typical Diameter: 10-50
Robust Background Global
Threshold Correction: 2
Lower & Upper Bounds: 0.001 - 1.0



This is getting close to where I’d like it, but the background is causing a real problem. Also, there is a tenancy to detect parallel FAs as a single object.

I also tried OtsuGlobal like you said, but I found Robust Background to be the best. I’d like to be able to detect all the long thin FAs, and also maybe be able to quantify the small ones, however distinguishing them from noise is tough.

Anything you could suggest that I might do differently?
Thanks,
Paul


FocalAdhesionDetection04.cp (5.29 KB)

Hi Paul,

The one suggesiton I have is to try to use CorrectIlluminationCalculate and CorrectIlluminationApply to remove the cell staining so you can detect the FAs better. In this case, the illumination correciton modules are not being used to handle illumination aberrations but as a means to estimate the background staining.

I’m attaching a pipeline that hopefully will set you in the right direction.

Regards,
-Mark
FocalAdhesionDetection04_MAB.cp (8.6 KB)