CRCNS: Waste-clearance flows in the brain measured using physics-informed neural network

CRCNS：使用物理信息神经网络测量大脑中的废物清除流量

• 批准号: 10613222
• 负责人: Douglas H Kelley
• 金额: $ 34.6万
• 依托单位: UNIVERSITY OF ROCHESTER
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-09-19 至 2027-06-30
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10613222
• 关键词: 3-Dimensional; Address; Alzheimer' s Disease; Artificial Intelligence; Biology; Brain; Brain imaging; Brain region; Cerebrospinal Fluid; Clinical; Communities; Contrast Media; Data; Engineering; Equation; Failure; Future; Health; High School Student; Human; Image; Intercellular Fluid; Link; Liquid substance; Magnetic Resonance Imaging; Measurement; Measures; Methods; Modality; Morphologic artifacts; Mus; Nature; Neurologic; Neurosciences; Noise; Pathologic; Pathologic Processes; Physics; Postdoctoral Fellow; Rehabilitation therapy; Research Personnel; Resolution; Science; Series; Sleep; Speed; Stroke; System; Techniques; Training; Traumatic Brain Injury; Variant; Work; contrast enhanced; data-driven model; experience; fluid flow; glymphatic flow; glymphatic function; glymphatic system; high risk; imaging modality; improved; in vivo; in vivo imaging; neural network; novel; outreach; particle; pressure; spatiotemporal; two-dimensional; two-photon; wasting

The brain’s transport system for cerebrospinal and interstitial fluid, the glymphatic system, was first described in 2012 by the Nedergaard team, operates primarily during sleep, and has been linked to pathological neurological conditions including Alzheimer’s disease, traumatic brain injury (TBI), and stroke. Obtaining quantitative measurements of glymphatic fluid velocity and pressure is crucial to understanding the function, failures, and potential rehabilitation of the glymphatic system. However, existing techniques for obtaining in vivo glymphatic velocities are limited to sparse measurements and specific regions, and pressure variation is essentially impossible to measure in vivo. We propose to quantify glymphatic flows from measurements of tracked particles and contrast agents using physics-informed neural networks (PINNs), which can infer velocity and pressure from sparse measurements and have not been used previously in neuroscience. We will adapt PINNs for three commonly-employed glymphatic imaging modalities: two-photon perivascular space imaging, transcranial whole-brain imaging, and dynamic contrast-enhanced magnetic resonance imaging (DCEMRI). For these modalities, each of which can probe different regions and scales of glymphatic flows, we will adapt the PINNs equations and artificial intelligence hyperparameters, evaluate the sensitivity of the approach to noise, spatiotemporal resolution, and imaging artifacts using synthetic data, and validate by comparing velocities inferred by PINNs to velocities from alternative techniques. Using PINNs will allow us to obtain in vivo velocity and pressure measurements of cerebrospinal fluid in previously unmeasured regions of the brain. Our collaborative team of neuroscientists, fluid dynamicists, and applied mathematicians includes the leaders who discovered the glymphatic system and invented PINNs. Moreover, we have extensive experience with all three imaging modalities and with velocity measurement (via automated particle tracking and front tracking) in glymphatic flows. This proposal seeks to reveal mechanisms by which the brain's transport system for cerebrospinal and interstitial fluid operates. Our novel velocity and pressure measurements of intracranial cerebral spinal fluid flows may demonstrate how improving sleep, the state during which the glymphatic system primarily operates, can counteract pathological processes related to glymphatic system failure including Alzheimer's disease.

脑内脑脊液和间质的运输系统 流体，即胶质淋巴系统，于2012年由Nedergaard团队首次描述，主要运作于 在睡眠中，并已与病理性神经系统疾病，包括阿尔茨海默氏病， 创伤性脑损伤（TBI）和中风。获得胶质淋巴液流速的定量测量 和压力是至关重要的了解功能，故障，和潜在的康复， 胶质淋巴系统然而，用于获得体内胶质淋巴速度的现有技术限于 稀疏测量和特定区域，并且压力变化基本上不可能测量 in vivo.我们建议量化胶质淋巴流的测量跟踪颗粒和对比度 使用物理信息神经网络（PINN）的代理，可以从 稀疏测量，并且以前在神经科学中没有使用过。我们将调整PINN， 三种常用的胶质淋巴成像模式：双光子血管周围空间成像， 经颅全脑成像和动态对比增强磁共振成像（DCEMRI）。 对于这些模式，每一种都可以探测不同的区域和胶质淋巴流的规模，我们 将适应PINNs方程和人工智能超参数，评估 方法噪声，时空分辨率，和成像伪影使用合成数据，并验证 将PINN推断的速度与替代技术的速度进行比较。使用PIN将允许 我们获得在体内的速度和压力测量的脑脊液在以前未测量 大脑的各个区域。我们的神经科学家、流体动力学家和应用科学家组成的合作团队 数学家包括发现胶质淋巴系统和发明PINN的领导者。 此外，我们在所有三种成像方式和速度方面都有丰富的经验， 测量（通过自动粒子跟踪和前端跟踪）胶质淋巴流。 这项提议旨在揭示脑的脑脊髓和神经元转运系统的机制。 间质液起作用。我们的新的速度和压力测量颅内脑脊髓 液体流动可以证明如何改善睡眠，在此期间，胶质淋巴系统 主要起作用，可以抵消与胶质淋巴系统衰竭有关的病理过程，包括 老年痴呆症