Optimization methods for characterization of single particles from light scattering patterns

Abstract

We address the inverse light-scattering problem for particles described by a several-parameters model, when experimental data are given as an angle-resolved lightscattering pattern (LSP). This problem is reformulated as an optimization (nonlinear regression) problem, for which two solution methods are proposed. The first one is based on standard gradient optimization method, but with careful choice of the starting point. The second method is based on precalculated database of theoretical LSPs, from which the closest one to an experimental LSP is selected for characterization. We tested both methods for characterization of polystyrene microspheres using a scanning flow cytometer (SFC).