A SCALABLE AND MODULAR PIPELINE FOR SYNTHESIZING AND RECONSTRUCTING BIOLOGICAL NETWORKS
Meher Niger, A Ashitkov, A Pillai, JD Wythe, G Chen, D Mayerich
Visualization of the 1000*1000*1000 dataset showing the extracted skeleton (in white) embedded within the vessel surface (in red). Progressively zoomed-in panels illustrate the correspondence between the skeletonized centerlines and the surrounding geometry.
Abstract
Sub-micrometer 3D imaging is essential for analyzing organ-scale tissue structures. Techniques such as light-sheet fluorescence microscopy (LSFM) have advanced whole-organ imaging, but introduce challenges in managing and processing whole-organ microscopy datasets. These challenges are compounded by a gap in expertise between biologists, instrument developers, and those creating algorithms and computing infrastructure. We introduce a modular open-source framework designed to promote collaboration and scalability. This framework includes modules for synthetic network generation, imaging simulation, segmentation, and centerline extraction. Post-simulation modules are built on the efficient OpenVDB data structure to facilitate development by reducing resource requirements.