Migration and division in cell monolayers on substrates with topological defects.

Collective movement and organization of cell monolayers are important for wound healing and tissue development. Recent experiments highlighted the importance of liquid crystal order within these layers, suggesting that +1 topological defects have a role in organizing tissue morphogenesis. We study fibroblast organization, motion, and proliferation on a substrate with micron-sized ridges that induce +1 and -1 topological defects using simulation and experiment. We model cells as self-propelled deformable ellipses that interact via a Gay-Berne potential. Unlike earlier work on other cell types, we see that density variation near defects is not explained by collective migration. We propose instead that fibroblasts have different division rates depending on their area and aspect ratio. This model captures key features of our previous experiments: the alignment quality worsens at high cell density and, at the center of the +1 defects, cells can adopt either highly anisotropic or primarily isotropic morphologies. Experiments performed with different ridge heights confirm a prediction of this model: Suppressing migration across ridges promotes higher cell density at the +1 defect. Our work enables a mechanism for tissue patterning using topological defects without relying on cell migration.

Deep Stroke-Based Sketched Symbol Reconstruction and Segmentation.

Hand-drawn objects usually consist of multiple semantically meaningful parts. In this article, we propose a neural network model that segments sketched symbols into stroke-level components. Our segmentation framework has two main elements: a fixed feature extractor and a multilayer perceptron (MLP) network that identifies a component based on the feature. As the feature extractor we utilize an encoder of a stroke-rnn, which is our newly proposed generative variational auto-encoder (VAE) model that reconstructs symbols on a stroke-by-stroke basis. Experiments show that a single encoder could be reused for segmenting multiple categories of sketched symbols with negligible effects on segmentation accuracies. Our segmentation scores surpass existing methodologies on an available small state-of-the-art dataset. Moreover, extensive evaluations on our newly annotated big dataset demonstrate that our framework obtains significantly better accuracies as compared to baseline models. We release the dataset to the community.