Precise characterization of nanometer-scale systems using interferometric scattering microscopy and Bayesian analysis

TL;DR

This paper introduces a Bayesian modeling approach for interferometric scattering microscopy that improves the precision of nanoscale system characterization by directly fitting the interferometric point spread function to data.

Contribution

It develops a rapid, parameterized model for interferometric images enabling Bayesian analysis with Hamiltonian Monte Carlo, enhancing measurement accuracy and data utilization.

Findings

Accurately determines 3D position and polarizability of nanoparticles.

Infers diffusion coefficients without mean-square displacement calculations.

Quantifies DNA ejection from individual viruses.

Abstract

Interferometric scattering microscopy (iSCAT) can image the dynamics of nanometer-scale systems. The typical approach to analyzing interferometric images involves intensive processing, which discards data and limits the precision of measurements. We demonstrate an alternative approach: modeling the interferometric point spread function (iPSF) and fitting this model to data within a Bayesian framework. This approach yields best-fit parameters, including the particle's three-dimensional position and polarizability, as well as uncertainties and correlations between these parameters. Building on recent work, we develop a model that is parameterized for rapid fitting. The model is designed to work with Hamiltonian Monte Carlo techniques that leverage automatic differentiation. We validate this approach by fitting the model to interferometric images of colloidal nanoparticles. We apply the method to track a diffusing particle in three dimensions, to directly infer the diffusion coefficient of a nanoparticle without calculating a mean-square displacement, and to quantify the ejection of DNA from an individual lambda phage virus, demonstrating that the approach can be used to infer both static and dynamic properties of nanoscale systems.