GPU-Accelerated Monte Carlo Photon Transport Simulation Platform

 描述（由申请人提供）：新型生物光子学技术以前所未有的速度不断涌现，以满足对准确、快速、无创或微创生理学定量的快速扩展的临床需求。对光子和生物组织之间复杂相互作用的透彻理解是这些技术的基础。在过去的5年里，我们的团队一直致力于开发计算效率高的蒙特卡罗（MC）方法，用于模拟复杂组织结构内的光传输。作为这项研究的结果，两个开源的模拟包-蒙特卡洛极端（MCX）和基于网格的蒙特卡洛（MMC）-已经开发和传播。他们现在正在推动全球许多光学实验室以及小型企业的研究，全球有超过240次引用，超过9，000次下载和近20，000次独立网络访问者。在本提案中，我们寻求进一步扩展，巩固和传播MCX和MMC，遵循NIH PA-14-156中的具体指南。将开发和实施用于建模宽视场照明和检测、快速断层重建和进一步速度加速的基于MC的新算法，以满足整个社区对开发下一代光学成像技术不断升级的需求。作为经验丰富的开源开发人员和维护人员，我们接受与用户社区互动的关键作用，并计划显着增强我们的软件用户体验，用户支持和系统的推广和培训。为了显著提高可用性和可扩展性，我们将开发基于Web的MCX/MMC，使我们的平台可以广泛访问，并开发分布式计算云，以快速处理大型复杂的多模态数据集。我们还将开发正式的培训课程和材料，以更好地教育研究人员，加强支持和反馈机制，并致力于通过我们的平台创建标准化的光学成像数据交换规范。实现这些目标不仅将使MCX/MMC成为最准确，高效和全面的光学建模平台之一，而且还将为我们的社区建立一个新的标准，用于开发创新的生物光子技术，探索复杂的生物系统，促进可重复的研究，并在广泛的研究社区之间建立有效的合作。

 DESCRIPTION (provided by applicant): Novel biophotonics techniques continue to emerge at an unprecedented pace to address the rapidly expanding clinical needs for accurate, fast, non- or minimally-invasive quantification of physiology. A thorough understanding of the complex interaction between photons and biological tissues is at the very foundation of these techniques. Over the past 5 years, our group has been dedicated to the development of computationally efficient Monte Carlo (MC) methods for modeling light transport inside complex tissue structures. As a result of this research, two open-source simulation packages - Monte Carlo extreme (MCX) and Mesh-based Monte Carlo (MMC) - have been developed and disseminated. They are now driving research in many optics labs across the world as well as in small businesses, attested to by over 240 citations, more than 9,000 downloads and nearly 20,000 unique web visitors worldwide. In this proposal, we seek to further extend, solidify, and disseminate MCX and MMC, following the specific guidelines in NIH PA-14-156. Novel MC- based algorithms for modeling wide-field illumination and detection, rapid tomographic reconstructions and further speed acceleration will be developed and implemented to meet the escalating needs throughout the community towards developing the next-generation optical imaging techniques. As experienced open-source developers and maintainers, we embrace the pivotal role of engagement with our user community and plan to significantly enhance our software user experience, user support and systematic outreach and training. Towards significantly improved usability and scalability, we will develop web-based MCX/MMC to make our platform widely accessible, and a distributed computing cloud to enable fast processing of large, complex multi-modal datasets. We will also develop formal training courses and materials to better educate researchers, strengthen the support and feedback mechanisms, and work to create a standardized optical imaging data- exchange specification through our platform. Accomplishing these goals will not only make MCX/MMC one of the most accurate, efficient and comprehensive optical modeling platforms available, but also will set a new standard in our community for developing innovative biophotonics techniques, exploring complex biological systems, facilitating reproducible research and forging efficient collaboration among a broad research community.