A Machine Learning Approach to Robustly Determine Orientation Fields and Analyze Defects in Active Nematics

Active nematics are dense systems of rodlike particles that consume energy to drive motion at the level of the individual particles. They are found in a variety of natural systems, such as biological tissues, and artificial materials such as suspensions of self-propelled colloidal particles or synthetic microswimmers. Active nematics have attracted significant attention in recent years because of their spectacular nonequilibrium collective spatiotemporal dynamics, which may enable applications in fields such as robotics, drug delivery, and materials science. The orientation field, which measures the direction and degree of alignment of the local nematic director, is a crucial characteristic of an active nematic and is essential for detecting and studying topological defects. However, determining the orientation field is a significant challenge in many experimental systems. Although orientation fields can be derived from images of active nematics using traditional imaging processing methods, the results and accuracy of such methods are highly sensitive to the settings of the algorithm. These settings must be tuned from image-to-image due to experimental noise, intrinsic noise of the imaging technology, and perturbations caused by changes in experimental conditions. This sensitivity currently limits the automatic analysis of active nematics. To overcome this limitation, we have developed a machine learning model for extracting reliable orientation fields from raw experimental images, which enables accurate automatic analysis of topological defects. Application of the algorithm to experimental data demonstrates that the approach is robust and highly generalizable to experimental settings that were unseen in the training data. These results suggest that it will be a useful tool for investigating active nematics, and the approach may be generalized to other active matter systems.