How to remove thin lines from the following image?

I am trying to isolate the circular droplets in the following image. However, I can't work out how to remove the thin lines of the chambers and if I just apply imfindcircles to the binarized image it does not provide me with what I need no matter how I tweak it. It keeps finding the empty chambers as circles (even though they are not complete circles) and it sometimes finds the odd shaped droplets and the droplets that are larger than the chambers (and therefore, not circular) as circles.

Once I isolate the circles I will track them over a few thousand images and monitor the mean gray level but I know how to do that.

Thanks in advance

[Walter Roberson]

What happens if you do a morphological thinning operation with a width that erases the thin lines but leaves the thicker lines present ?

[Daniel Powell]

Thanks, that's helpful.

It solves the part about getting rid of the chamber lines. However, the bigger droplets that aren't circular such as the one pictured below are still detected as "circles" by imfindcircles and if I lower the sensitivity so that these aren't detected then some of the other "genuine circles" are also excluded.

Answer 1

[Matt J]

Perhaps

bwpropfilt(bW,'Eccentricity',____)

Answer 2

[Steve Eddins]

Try using the 3rd output of imfindcircles (metric) to distinguish the bigger droplets that aren't really circles.