Blast Counts in Bone Marrow Aspirate Smears Analysis Using the Poisson Probability Function, Bayes Theorem, and Information Theory

Counts of cells or other phenomena observed through a microscope are numeric observations and, as such, are subject to mathematical and statistical analyses. For example, the Poisson probability function provides the probability of observing a particular number of blasts in a bone marrow aspirate, given an underlying true fraction of blasts present and a particular number of cells evaluated. Furthermore, using the Poisson function, Bayes theorem can provide the probabilities of specific categories of refractory anemia, given a number of observed blasts in a specific total of cells evaluated. Herein, I introduce and demonstrate these mathematical functions for the analysis of counts of blasts in marrow aspirates and explore the uncertainty that naturally arises when counts of blasts are near cut points used to separate the categories of refractory anemia without excess blasts, refractory anemia with excess blasts, and acute leukemia.