cristina-mt
commented
3 months ago The bounding box is built based on the minimum and maximum values for (x,y) obtained from the circular mask. Then the background correction using the corner is the mean intensity from the four corners in the bounding box, with each corner being 3x3 pixel area. This is generally a good approach to correct the local background for each vesicle, however there a few scenarios in which is not recommended:
1) there's too many vesicles (or fluorescent objects) in your image, likely causing for other vesicles to "invade" the bounding box corners
2) the fluorescence signal is very different between the corners of the bounding box.
3) the vesicles are very small, causing the corner area (3x3 box) of the bounding box to overlap with part of the vesicle.
Hi,
I've noticed that in the test data for basic encapsulation analysis using DisGUVery, the CHT method is used for vesicle detection, followed by the encapsulation module with 'B.box corner' for background correction. However, since the CHT method identifies vesicles as circles without explicit bounding boxes, could you clarify how the 'B.box corner' background correction is applied in this scenario? Is there an underlying process that converts circular detections to bounding boxes for background correction, or is there another recommended approach for this case?
Thank you for your insights.
Best regards, Sree