Mutual information of input and output of Spatial pooler

Hi everyone
I saw in some Numenta codes that entropy was implemented to check whether the activity of the columns was the same or not.Now my question is whether it is possible to calculate, for example, the mutual information between the input and output of the
Spatial pooler algorithm?

I know we need distributions to calculate mutual information, but I do not know whether we have distributions here or not (for example in the MNist classification with Spatial pooler algorithm)
I am waiting for your comments