High Throughput Photopatterning and Interactive Manipulation of Microparticles and Microorganisms.

Abstract

Recent advances in soft material microfabrication technologies are enabling wide-ranging studies of cellular and organism behavior in vitro; however, these methods are generally time-consuming, challenging to implement by non-experts, are limited to planar features, and cannot be reconfigured within live environments. As a result, it is not possible to manufacture realistic artificial tissue constructs, nor to perform dynamic experimentation with model organisms. This thesis describes an integrated hardware and software platform, based on micro-scale light shaping and high-speed machine vision algorithms that enables real-time, dynamic photo-patterning in response to microscale environmental changes. An optofluidic lithography system designed for the purpose of in-flow polymerization of hydrogel microstructures achieved diffraction limited resolution (r = 0.7µm) with a maximum distortion of the projection of 160nm. This enables continuous production of poly(ethylene-glycol) diacrylate(PEG-DA) microparticles (20-100μm, CoV5-15%). A new pillared microfluidic device design increased throughput up to 1500-fold, capable of synthesizing 2.5×〖10〗^6 particles per minute. Biocompatibility of hydrogels was validated for model organism C. elegans, and hepatocytes. Dynamic assays where structures were built during live culture affirm that proximity of pillared structures increased the swimming speed of C. elegans and showed that worm behavior can be influenced by sequential photopatterning of free-floating structures. A software architecture was designed to enable use of machine vision to in flow, by photopolymer encapsulation in response to image-based decision events. We then evaluated the sensitivity, specificity, RMSE and computational time of candidate machine vision algorithms, and find the Speeded Up Robust Feature (SURF) method was the most robust though Thresholding was 3 orders of magnitude faster than SURF. Using this capability, we sorted poly(styrene) micro particles by size via selective encapsulation (TPR=100% and SPC=99.999%, Mean error 4.7 pixels); and print patterns of hepatocyte aggregates with single cell resolution (<20µm) onto polymer substrates. Last, the thesis describes the design and testing of a six-axis robotic dynamic lithography system for patterning large area curved surfaces. Looking forward, platforms combining micro- and nanofabrication processes with image-driven artificial intelligence algorithms could widely expand capabilities for scalable biofabrication and automation of science, including for custom fabrication of cell-based assays and in vitro organ mimics.