AI aided microfluidic auto continuous separation and sorting of microspheres simultaneously based on their morphological and optical properties

Abstract

We report a novel method of continuous online microsphere/particle separation and sorting that integrated techniques of microsphere-imaging, microfluidic chip, moving micro-object super-resolution analysis, and AI image recognition into a simple system. It is the first method with the ability to determine the size of moving micro-objects with nanoscale resolution and separate/sort them simultaneously according to their size, shape, and refractive index. In the experiment, the separation and sorting of polystyrene, silica, and PMMA microspheres from the mixture solutions of several different microspheres using a microfluidic device demonstrated that the sorting was accurate and highly efficient and could classify micro-objects with a nanoscale size difference of ∼ 10 nm and a refractive index difference of 10−5. The method is suitable for coupling with various micro-object separation/sorting instruments for different micro-object separations. It can also divide micro-objects either with or without coating.