4D Imaging of Spatially and Temporally Dynamic Biophysical Processes using Sparse Data Methods

提出的发现者计划将建立利用时空数据稀疏性的新的成像物理技术，并相应地提高我们对特定快速动态生理过程的生物物理理解。获取高质量的图像--即高信噪比、对比度和空间分辨率--通常需要大量数据，因此似乎与获得准确表征动态系统所需的高时间分辨率数据不一致。最近，利用空间和时间冗余(即数据稀疏性)进行数据采样和图像重建的新策略提高了我们获得高质量定量地图的能力，这些地图准确地模拟了随时间快速变化的生物物理测量。虽然在这一领域已经取得了长足的进步，但许多高度动态的生物物理过程仍然没有得到研究，因为目前的工具不能提供足够的空间和时间分辨率来准确地描述它们。 此外，在研究个体而不是平均总体时，这一问题变得更加具有挑战性，因为获取和分析数据的最佳策略在一定程度上是由该精确数据集中的时空特征驱动的。关键是，虽然新的数据采样技术可能会导致高加速系数，但物理学家必须对简单地试图走得更快保持警惕，因为他们可以走得更快。“正确的”数据采样和重建方案是将测量数据与感兴趣的实际生物物理特性最佳地连接起来的方案。确定什么是正确的战略本身可能是一个巨大的挑战。因此，模拟和与生物学的相关性至关重要。 该计划连贯地将新的数据采集技术、图像重建技术和分析工具的基础研究结合在一起，以改进人体时空动态系统的图像表征。该计划内的所有项目都将建立在假设生成理论模拟的基础上，这些模拟随后将通知和指导用于获取和/或分析经验数据的新技术的开发，并通过与潜在生物物理属性的关联进行验证。 这项工作将应用于似乎不同的生理过程(例如，动态对比增强和功能神经成像)和成像技术(例如，MRI和脑磁图)。然而，这个项目中的所有研究都有一个共同点，那就是它需要4D数据，这些数据必须被获取和分析，以便它可以与个体感兴趣的生理参数相联系。这样做将建立在成像物理和数据压缩领域的新的数学和物理方法的基础上，并将直接来自我之前在NSERC发现基金中进行的研究。

The proposed Discovery program will establish new imaging physics technologies utilizing spatiotemporal data sparsity, and correspondingly improve our biophysical understanding of specific rapidly dynamic physiological processes. Acquiring high quality - i.e., high signal-to-noise, contrast and spatial resolution - images typically requires large amounts of data, and is therefore seemingly at odds with obtaining the high temporal resolution data needed to accurately characterize dynamic systems. Recently, novel strategies for both data sampling and image reconstruction that take advantage of spatial and temporal redundancies (i.e. data sparsity), have improved our ability to acquire high quality quantitative maps that accurately model biophysical measurements that change quickly in time. While great strides have been made in this area, many highly dynamic biophysical processes remain unstudied, as the current tools do not provide sufficient spatial and temporal resolution to accurately characterize them. Furthermore, this problem becomes even more challenging when studying individuals, rather than averaged populations, as the optimal strategy to both acquire and analyze the data is in part driven by the spatiotemporal features within that exact data set. Critically, while novel data sampling techniques can lead to high acceleration factors, physicists must be wary of simply trying to go faster because they can go faster. The “correct” data sampling and reconstruction scheme is the one that optimally connects the measured data with the actual biophysical property of interest. Determining what the correct strategy is can be a significant challenge in its own right. Hence, simulation and correlation to biology are critical. This program coherently brings together basic research into new data acquisition technologies, image reconstruction techniques, and analysis tools for improving image characterization of spatiotemporally dynamic systems in the human body. All projects within this program will build from hypothesis generating theoretical simulations that will subsequently inform and guide the development of new technologies for the acquisition and/or analysis of empirical data, with validation through correlation to the underlying biophysical properties. This work will be applied to a seemingly diverse spectrum of physiologic processes (e.g., Dynamic Contrast Enhancement and Functional Neuroimaging) and imaging technologies (e.g., MRI and MEG). However all research in this program has the common thread that it requires 4D data that must be acquired and analyzed such that it can be connected to a physiological parameter of interest in an individual. Doing so will build off of novel mathematical and physics approaches from the fields of imaging physics and data compression, and will flow directly from the research performed in my previous NSERC Discovery grant.