Large‐scale volume data analysis for meso‐ and macroscopic physiological modeling

Modeling is increasingly important to biological investigations, from basic cell biology to cognitive neuroscience. For many years, however, models and simulations have been severely limited by a lack of spatial, i.e. anatomical, realism. Simulation methods that incorporate highly detailed anatomy are expensive and complex, but developers are targeting ever increasing 3‐D spatial realism and have made major advances in recent years. Thus there is a renewed need for quantitative 3‐D anatomical data at subcellular to tissue scales, with many remaining challenges: (1) Automating detailed analyses from varied imaging modalities; (2) Handling potentially enormous datasets; (3) Correcting artifacts introduced by tissue preparation and imaging method; (4) Annotating datasets to include model parameters, e.g., molecular constituents and physical properties of intra‐ and extracellular spaces and membrane compartments; (5) Production of surface and volume meshes that include annotated properties; and (6) Formatting of meshes and properties for use with one or more simulation methods. We are developing PSC Volume Browser , DReAMM , and MCell software to address all of these challenges, with special emphasis on arbitrarily large volumetric datasets and Monte Carlo diffusion‐reaction simulation methods. We will present examples ranging from semi‐automated analysis of cardiac time series from micro‐CT imaging, to nonlinear registration and alignment of large‐scale serial EM sections, surface mesh generation using volume segmentation and computer‐aided design approaches, and quantitative simulation of microphysiological events such as calcium‐dependent neurotransmitter release from nerve terminal active zones. Supported by NIH P41 RR06009, R01 GM068630, NLM N01‐LM‐9‐3531, and NIH/NSF 0234002.