Simulation-guided regression approach for estimating the size distribution of nanoparticles with dynamic light scattering data

This article presents a simulation-guided regression approach for estimating the size distribution of nanoparticles from Dynamic Light Scattering (DLS) measurements. The properties and functionalities exhibited by nanoparticles often depend on their sizes, so the precise quantification of the sizes is important for characterizing and monitoring the quality of a nanoparticle synthesis process. The state-of-the-art method used in the size quantification from DLS measurements is the CONTIN, which is based on a computationally ineffective numerical inversion. We propose a new approach that avoids the numerical inversion by reformulating the problem into a regularized regression problem, with the basis functions being generated by a computer simulation of DLS measurements. For many simulation studies and one real data study, our method outperformed the CONTIN in terms of estimation accuracy and computational efficiency.