How do I mask out the largest object and everything below it?

Hello,

I am trying to analyse optical coherence tomography images of the retina. I would like to automatically count hyperfeflective dots in the vitreous (the area above the retina) and obtain an average intensity for the vitreous.

I have binarized the images (see binarized.jpg attached) and would like to mask out the largest object, and everything below it, since there are also some hyperreflective dots (of a similar size) in that region of the image which I do not want to count. In addition I must also only consider the region above the retina when evaluating the average intensity.

I am unsure how to execute this and any help would be much appreciated.

Thank you in advance.