Learning-Based Cell Injection Control for Precise Drop-on-Demand Cell Printing

Abstract

Drop-on-demand (DOD) printing is widely used in bioprinting for tissue engineering because of little damage to cell viability and cost-effectiveness. However, satellite droplets may be generated during printing, deviating cells from the desired position and affecting printing position accuracy. Current control on cell injection in DOD printing is primarily based on trial-and-error process, which is time-consuming and inflexible. In this paper, a novel machine learning technology based on Learning-based Cell Injection Control (LCIC) approach is demonstrated for effective DOD printing control while eliminating satellite droplets automatically. The LCIC approach includes a specific computational fluid dynamics (CFD) simulation model of piezoelectric DOD print-head considering inverse piezoelectric effect, which is used instead of repetitive experiments to collect data, and a multilayer perceptron (MLP) network trained by simulation data based on artificial neural network algorithm, using the well-known classification performance of MLP to optimize DOD printing parameters automatically. The test accuracy of the LCIC method was 90%. With the validation of LCIC method by experiments, satellite droplets from piezoelectric DOD printing are reduced significantly, improving the printing efficiency drastically to satisfy requirements of manufacturing precision for printing complex artificial tissues. The LCIC method can be further used to optimize the structure of DOD print-head and cell behaviors.

Machine Assisted Experimentation of Extrusion-Based Bioprinting Systems

“…The inclusion of ML in 3D bioprinting is relatively new, although unique contributions have been made thus far. Shi et al implemented a multilayer perceptron-based artificial neural network trained with computational fluid dynamics simulations of droplet formation and flow behavior to predict classification-based droplet behavior using voltage, nozzle diameter, bioink surface tension, and bioink viscosity input parameters for a drop-on-demand bioprinting system [8]. Experimental validation of six different input parameter combinations that were predicted to produce single, satellite, or no droplets confirmed for each case that experimental results matched with droplet formation predictions.…”

Printability and Cell Viability in Extrusion-Based Bioprinting from Experimental, Computational, and Machine Learning Views

“…Shi et al [ 145 , 146 ] employed a multiobjective optimization method and artificial neural network with computational fluid dynamics to simulate droplet formation and flow behaviour in drop-on-demand printing. Printing Silicone elastomer via freeform reversible embedding (FRE) is challenging due to depositing a Newtonian ink within a Bingham plastic support.…”

Engineering Tissue Fabrication With Machine Intelligence: Generating a Blueprint for Regeneration

“…Considering the complexity of these additive manufacturing techniques and their potential application to tissue fabrication, it is not surprising to find various methods ranging from biologically inspired ones such as genetic algorithms (GA, which mimic the process of natural selection, de Castro, 2007;Paszkowicz, 2009) to statistical and probabilistic algorithms. They could be grouped as (i) optimal design methods with DOE and its variants such as Taguchi method (Mohamed et al, 2016;Scaffaro et al, 2017;Yousefi et al, 2019), (ii) optimization with population-based methods (Rahmani-Monfared et al, 2013;Asadi-Eydivand et al, 2016;Rao and Rai, 2016;Heljak et al, 2017;Abdollahi et al, 2018), and (iii) problem specific approaches often facilitated by ML (Cheheltani et al, 2012;Farzadi et al, 2015;Tiwari et al, 2015;Langelaar, 2016;Querido et al, 2017;Saadlaoui et al, 2017;Gholami et al, 2018;Shi et al, 2018Shi et al, , 2019Menon et al, 2019;Zhang et al, 2019;Zohdi, 2019). Despite the abovementioned progress, there is still a significant gap in combining data-to-blueprint translation (Figure 1), which is the process of extracting information from data and incorporating the information in a blueprint file [e.g., CAD (computer-aided design) files] for 3DBP.…”