Binary methods for the microdosimetric analysis of cell survival data from alpha-particle irradiation

A new type of alpha-particle irradiator allows survival of each cell to be observed individually along with the size and shape of its nucleus and the positions of the hits it receives. This paper discusses methods of data analysis that can utilize these additional data. Using idealizations of the cell nucleus geometry (i.e., spheres, ellipsoids), the path length (l), energy deposited (e), and specific energy (z) has been determined on a cell-by-cell basis for 772 cells all subjected to the same fluence. Each cell is regarded as a Bernoulli trial with a different probability for success (colony formation). For the survival expectation, A exp(-z/z(o)), the values of A and z(o) are chosen to maximize the likelihood for the observed outcome. Similar results are presented using the alternate functional forms A exp(-e/e(o)) and A exp(-l/l(o)). With these parameter values, the goodness of fit is also evaluated using a chi-square method with variances given by the binary (Bernoulli) methods. A further purpose of the paper is to assess the validity of the microdosimetric computations that would have had to be made if these individual cell-by-cell experimental measurements were not available or were incomplete.