Based on the above, we can say that:
a = A * (selection probability for the diseased)
b = B * (selection probability for the non-diseased)
If the selection probabilities are known, then each observed value; 'a' and 'b' can be divided by these probabilities to estimate the number of diseased and non-diseased within the population. Converting this into a proportion leads to an adjusted prevalence estimate.
Alternatively, you can take the observed odds of disease and multiply this by the inverse of the selection odds (the ratio of selection probabilities for the diseased and non-diseased) to produce an adjusted odds of disease. This can be converted back to a proportion, if you wish.