Online Multivariate Statistical Neural Imaging Data Analysis

描述（由申请人提供）：近年来，成像技术的爆炸式发展使我们非常接近于实现利用单神经元分辨率对大脑功能进行成像的目标：使用标记技术如表达的报告基因和大量染料加载，结合成像方法如快速帧CCD、光电二极管阵列和传统以及多光子共聚焦成像，我们现在可以在细胞培养、脑切片制备甚至完整动物的水平上，以单细胞分辨率可视化复杂的神经相互作用。 在我们自己的工作中，我们发现同样的技术与现代计算机相结合，使我们能够更快地收集大量信息，而不是使用标准的工作流模型来处理和理解它。 此外，迫切需要采用自动数据库存储和分析结果分类的系统生物学策略。 我们建议解决这些问题和其他问题，重点技术开发建议，解决问题/需求为基础的标准所解决的神经技术研究，开发和增强计划。 PI具有实验和计算建模背景，而co-PI具有神经成像数据分析背景，为该项目带来了补充技能。 我们将开发一个通用的框架，用于数据收集，元标记和基于图像的时间序列存储。 我们将开发和实施算法，将允许在线计算的近似或精确的主成分和多元光谱特性在实验操作。 我们将实现典型相关分析（CCA）和其他技术来比较多个数据集。 通过实施多变量方法的比率成像数据的分析，我们将奠定基础，将这些多变量的技术和工具，以全光谱为基础的数据收集。 . 公共卫生相关性：生理成像越来越多地帮助我们了解大脑在健康和疾病中的工作方式。 这项工作与公共卫生有关，因为它将提供必要的工具，使生理成像更加有效和灵敏。 此外，它将为使用生理成像作为高通量工具铺平道路。

DESCRIPTION (provided by applicant): In recent years, an explosion of imaging technology has brought us tantalizingly close to achieving the goal of imaging brain function with single neuron resolution: Using labeling techniques such as expressed reporters and bulk dye loading combined with imaging methods such as fast frame CCD, photodiode array, and traditional as well as multiphoton confocal imaging, we can now visualize complicated neural interactions at the level of cell culture, the brain slice preparation and even the intact animal, with single cell resolution. In our own work, we have found that this same technology paired with modern computers has resulted in the ability to collect masses of information far more quickly than we can process and understand it using standard workflow models. Additionally, there is a pressing need to adopt the systems biology strategy of automated database storage and classification of analysis results. We propose to address these and other issues with a focused technology development proposal which adreses the problem/need based criteria addressed by the Neurotechnology Research, Development and Enhancement Program. The PI, with an experimental and computational modeling background, and the co-PI, with a neural imaging data analysis background, bring complementary skills to this project. We will develop a common framework for data collection, meta-tagging and storage of image based time series. We will develop and implement algorithms that will allow for the on-line calculation of approximate or exact principal components and multivariate spectral characteristics during experimental manipulations. We will implement Canonical Correlation Analysis (CCA) and other techniques for the comparison of multiple datasets. By implementing multivariate methods for the analysis of ratiometric imaging data, we will lay the groundwork for extending these multivariate techniques and tools to full optical spectrum based-data collection. . PUBLIC HEALTH RELEVANCE: Increasingly, physiological imaging is helping us to understand how the brain works in health and disease. This work is relevant to public health because it will provide necessary tools for making physiological imaging more efficient and sensitive. In addition, it wil pave the way for using physiological imaging as a high-throughput tool.