Original title: ToSkA: Topological Skeleton Analysis for Network-Based Shape Representation and Evaluation of Objects from Cells to Death Stars
Authors: Allyson Quinn Ryan, Carl D. Modese
The article introduces “ToSkA: Topological Skeleton Analysis,” a method devised for analyzing complex object shapes through network representations. Traditional shape descriptors might lack adequacy when assessing intricate shapes. ToSkA creates object profiles using network science metrics, spatial embeddings, and global neighborhood principles, facilitating comprehensive shape evaluation and classification. Notably, it captures object evolution over time and distinguishes significant shape variations between experiments. The method accounts for absolute spatial features like branch lengths, enhancing classification sensitivity. Additionally, as topology defines system identity, ToSkA detects segmentation errors altering object topology by observing changes in network cycles. This approach’s flexibility and in-depth shape profiling suit biological and physical settings, enabling precise configuration analysis crucial for downstream processes in systems demanding robustness and precision.
Original article: https://arxiv.org/abs/2311.10539