High-resolution Nanoparticle Sizing With Maximum A posteriori Nanoparticle Tracking Analysis (MANTA)
Kevin Silmore, Massachusetts Institute of Technology
Unbiased estimation of nanoparticle size distributions remains a challenge affecting numerous fields, including nanotechnology and biopharmaceuticals. Current methods for particle sizing, such as dynamic light scattering, analytical ultracentrifugation and field-flow fractionation, can suffer from a combination of intensity biases, difficult sample preparation, insufficient sampling and ill-posed data analysis. The last issue, ill-posed governing equations for these particle sizing techniques, guarantees that noise in the data will make reproducible measurements impossible without aggressive regularization. We have developed a new Bayesian method, Maximum A posteriori Nanoparticle Tracking Analysis (MANTA), for estimating the size distributions of nanoparticle samples from high-throughput, single-particle tracking experiments. In this method, unbiased statistical models are derived for two observable quantities in a typical nanoparticle trajectory (mean square displacement and the trajectory length), finite elements are used to discretize the particle size distribution, the weights of the finite elements are found via MAP estimation, and robust cross validation is applied to mildly regularize solutions to the MAP problem. MANTA infers nanoparticle size distributions with unprecedented resolution. The precision of the method is assessed by performing extensive Brownian dynamics simulations with synthetic particle size distributions that are known exactly. In experimental testing with gold nanoparticles, MANTA is capable of reliably resolving disparate populations of particles with relative variation in size as small as 30 percent.
Abstract Author(s): Kevin Silmore, Xun Gong, Michael Strano, James Swan