A Computational Facility for Neuroimaging Research

描述(申请人提供)：使用磁共振成像、磁共振光谱学、磁共振显微镜和密集阵列电记录的神经成像研究涉及获取非常大的数据结构，然后必须通过一系列计算密集型后处理任务将其转换为有意义的信息。这个项目是由杜克大学和北卡罗来纳大学的NIH支持的研究人员联合提交的，他们主要使用神经成像方法。研究人员代表了支持这些机构神经成像研究的四个跨学科中心：(1)杜克-北卡罗来纳大学脑成像和分析中心(由两所大学联合资助)，(2)活体显微镜中心(NIH国家研究资源中心)，(3)高级磁共振发展中心，(4)认知神经科学中心。这些中心的主任和他们的合作科学家发现，由于采集方法的改进和硬件的改进而产生的海量数据，使得及时全面分析数据变得越来越困难。一些重要的计算过程是不可行的，例如创建3D化学位移图像。其他常规过程，如图像体积时间序列的3D空间归一化，需要在我们的单处理器工作站上进行数小时的处理。我们正在申请资金购买一台高性能多处理器计算机，以满足这些和其他处理要求。 我们的提案主要是利用杜克大学最近安装的IBM存储区域网络(SAN)，该存储区域网络具有大容量的多TB存储容量，供其高科技中心使用。拟议的多处理器系统将通过光纤通道适配器直接连接到这个SAN。因此，拟议的计算资源将充分实现SAN对我们研究人员小组的好处。我们的建议还得益于纳入的中心在科学努力、分析软件开发和大型研究仪器管理方面已经存在的密切合作。

DESCRIPTION (provided by applicant): Neuroimaging research using magnetic resonance imaging, magnetic resonance spectroscopy, magnetic resonance microscopy, and dense array electrical recording involves the acquisition of very large data structures that then must be transformed through a series of compute-intensive post-processing tasks into meaningful information. This project is jointly submitted by NIH-supported investigators at Duke University and the University of North Carolina who primarily use neuroimaging methods. The investigators are representative of four interdisciplinary centers that support neuroimaging research at these institutions: (1) The Duke-UNC Brain Imaging and Analysis Center (jointly funded by both universities), (2) the Center for In- Vivo Microscopy (a NIH National Research Resource Center), (3) the Center for Advanced Magnetic Resonance Development, and (4) the Center for Cognitive Neuroscience. The directors of these centers and their collaborating scientists have found that the sheer volume of data resulting from improved acquisition methods and improved hardware has made it increasingly difficult to fully analyze data in a timely manner. Some important computational processes are unfeasible, such as the creation of 3D chemical shift images. Other routine processes such as 3D spatial normalization of a time series of image volumes require many hours of processing on our single processor workstations. We are requesting funds to purchase a high performance multiprocessor computer to facilitate these and other processing requirements. Our proposal trades heavily upon the recent installation at Duke University of an IBM Storage Area Network (SAN) with capacious multi-terabyte storage capacity for the use of its high technology centers. The proposed multiprocessor system would be connected directly to this SAN via a fiber channel adapter. Thus, the benefits of the SAN to our group of investigators would be fully realized by the proposed computational resource. Our proposal also benefits from the close collaboration that already exists among the included centers in scientific endeavors, in the development of analytical software, and in the management of large research instruments.