Best practices to determine co-expression?


I have several gene-gene pairs and I would like to classify them as either co-expressed or mutually exclusive. I created a bunch of scatter plots with expression of one gene vs the other plotted for all cells in a cluster of interest. I naively expected the plots to look unambiguous: all points on the axes = mutually exclusive, all points in the middle = co-expressed. but I realized that even in “positive control” cases when the two genes are expected to be simultaneously expressed, there are always cells that express only one gene and not the other. are these dropouts? what would be the best way to go about analyzing co-expression in this case?

thanks in advance!