Extract well picture

[demon]

Hi KNIME community,

Here is my problem: I have a microtiter plate with cell aggregations:

The images are taken manually and all are in the same focal plane, but the well center varies from picture to picture. I want to extract the aggregates and measure the aggregate area, however, the biggest issue I’m facing currently is when the cell aggregates are stuck to the well wall, i.e. I cannot distinguish the well border from the aggregate anymore. What is the best way to solve this issue, i.e. automatically recognize the well and extract the well picture, afterwards applying the connected component analysis?

Thanks for your help.

[gab1one]

Hi @demon,
If I understand you correctly, you want to isolate the region I marked in yellow below? And then run measurements on only the marked region?

And your thresholding fails when the dark stuff is at the border of the well?
You can try applying the Convex Hull operation to the extracted mask, this should clean up the mask.

Convex Hull

best,
Gabriel