User-friendly analysis of droplet array images

Abstract

Droplet-based experimental platforms allow researchers to perform massive parallelization and high-throughput studies, such as single-cell experiments. Even though there are various options of image analysis software to evaluate the experiment, selecting the right tools require experience and is time consuming. Experts and sophisticated workflow are required to perform the analysis, especially to detect the droplets and analyze their content. There is need for user-friendly droplet analysis pipelines that can be adapted in laboratories with minimum learning curve. Here, we provide a user-friendly workflow for image-based droplet analysis. The workflow comprises of a) CellProfiler-based image-analysis pipeline and b) accompanied with web application that simplifies the analysis and visualization of the droplet-based experiment. We construct necessary modules in CellProfiler (CP) to detect droplets and export the results into our web application. Using the web application, we are able to process and provide basic profiles of the droplet experiment (droplet sizes, droplet signals, sizes-signals plot, and strip plot for each label/condition). We also add a specific module for growth heterogeneity studies in bacteria populations that includes single cell viability analysis and probability distribution of minimum inhibition concentration (MIC) values in population. Our pipeline is usable for both poly- and monodisperse droplet emulsions.

Vision-Based Performance Analysis of an Active Microfluidic Droplet Generation System Using Droplet Images

“…Size [32] Eccentricity [33] Gradient [34] Concavity [35] Projected Surface Area [36] Sphericity Index [37] Symmetry [38] Roundness [39] Border [40] Radius [41] Membrane Surface Area [42] Contrast Variations [30] Shape [43] Elongation [44] Saturation [45] Volume [38] Sphericity Coefficient [46] Form factor [47] Precise droplet production and manipulation are critical in droplet-based bioassays. Droplet-based analytical devices have demonstrated promising results, but most systems are trained to generate predefined droplets, and the devices are incapable of adapting to performing complex bioassays which require step-by-step processing.…”