A high-throughput method to characterize membrane viscosity of flowing microcapsules

Microcapsules have many industrial applications and also serve as a widely used mechanical model of living biological cells. Characterizing the viscosity and elasticity of capsules at a high-throughput rate has been a classical challenge, since this is a time-consuming process in which one needs to fit the time-dependent capsule deformation to theoretical predictions. In the present study, we develop a novel efficient method, by integrating a deep convolutional neural network with a high-fidelity mechanistic capsule model, to predict the membrane viscosity and elasticity of a microcapsule from its dynamic deformation when flowing in a branched microchannel. Compared with a conventional inverse method, the present approach can increase the prediction throughput rate by five orders of magnitude while maintaining the same level of prediction accuracy. We also demonstrate that the present approach can deal with capsules with large deformation in inertial flows.

Automated summary

In this study, a novel efficient method combining a deep convolutional neural network with a high-fidelity mechanistic capsule model is proposed to predict the membrane viscosity and elasticity of microcapsules flowing in a branched microchannel. Compared with traditional methods, this approach significantly increases prediction throughput rate while maintaining accuracy, and can handle capsules with large deformation in inertial flows.