Determine the data of the graphics part in the web

Hello everyone,
I encountered a problem in development today. I have a Datatable of a webpage, which contains circles. The interesting thing is that they use hollow and solid to represent read or unread. I want to take out those unread data. I have encountered a bottleneck , Data capture does not show this picture. Because he has thousands of pages, I want to use Find Children to find the judgment selector, and then get his index. The fact is that although I succeeded, he runs really slow, 100 pieces of data takes five minutes, it is difficult to accept Is there a better method for this, I would like to refer to it.

Can you answer my doubts? If anyone does, I really appreciate it.