Sensitive detection and estimation of particle non-sphericity from the complex Fourier spectrum of its light-scattering profile

摘要

We develop a fast method to estimate the non-sphericity of arbitrary-shaped particles from the complex Fourier spectrum of its light-scattering profile (LSP), measured with the scanning flow cytometer (SFC). We show that previously used amplitude spectrum is not sufficiently sensitive to the non-sphericity and extensively study the phase of the spectral peak for spheroids in the framework of the Rayleigh-Gans-Debye (RGD) approximation. Based on this analysis we construct a new spectral parameter P - the weighted deviation of the complex spectrum around the peak from that for an equivalent sphere determined by the previously published spectral characterization method for spheres (SCMS). We also propose a geometric indicator of non-sphericity eta as the relative volume difference from that of the best-fit sphere. These two new parameters apply to particles of arbitrary shape and strongly correlate with each other for rigorously simulated LSPs for spheroids and biconcave disks in a wide range of sizes, refractive indices, and orientations. This correlation is the core of the new method, allowing one to provide both the estimate and the confidence range of eta from the experimental value of P. The method is both sensitive and specific to small non-sphericity. For instance, the median error of estimated aspect ratio for simulated LSPs of spheroids is 0.024. We test the resulting algorithm on the real experimental measurements of milk fat globules and red blood cells (RBCs) during the spherization process. These results raise a question about the actual shape of a spherized RBC in the flow inside the SFC. The applicability domain of the method is determined mainly by that of the SCMS and includes biological objects with sizes larger than 7 wavelengths in the liquid host medium. Moreover, we briefly discuss the potential extension of the method to larger refractive indices. (C) 2019 Elsevier Ltd. All rights reserved.