Functional MRI Method Development

功能性 MRI 方法开发

• 批准号: 10266587
• 负责人: Peter Bandettini
• 金额: $ 240.96万
• 依托单位: NATIONAL INSTITUTE OF MENTAL HEALTH
• 依托单位国家: 美国
• 资助国家: 美国
• 起止时间: 至
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10266587
• 关键词: 3-Dimensional; Address; Affective; Anatomy; Animal Model; Anterior; Area; Attention; Auditory; Auditory area; Behavioral; Blood; Brain; Brain Mapping; Brain region; COVID-19; Cognition; Collaborations; Communities; Complex; Computer software; Conscious; Data; Data Analyses; Data Collection; Development; Dimensions; Face; Functional Imaging; Functional Magnetic Resonance Imaging; Future; Goals; Healthcare; Human; Image; Individual; Inpatients; Intelligence; International; Laboratories; Letters; Light; Link; Liquid substance; Magnetic Resonance Imaging; Manuscripts; Mental Depression; Mental disorders; Methods; Modeling; Mood Disorders; Motion; Motor; Nature; Neurobiology; Neurons; Neurosciences; Newsletter; Noise; Nuts; Output; Paper; Pattern; Perfusion; Personality; Phenotype; Physiologic pulse; Physiology; Pial Veins; Prefrontal Cortex; Preparation; Protocols documentation; Publishing; Questionnaires; Research; Resolution; Rest; Safety; Sampling; Scanning; Sensory; Sequence Analysis; Series; Short-Term Memory; Signal Transduction; Sleep; Specific qualifier value; Specificity; Stimulus; Structure; Techniques; Testing; Time; Training; United States National Institutes of Health; Veins; Visual; Wakefulness; Work; base; behavioral phenotyping; behavioral study; brain behavior; cognitive ability; commune; data acquisition; design; hackathon; healthy volunteer; high resolution imaging; imaging modality; improved; insight; large scale data; method development; movie; multimodality; multisensory; neural patterning; neuroimaging; novel; novel strategies; open source; predictive modeling; relating to nervous system; response; social; tool; trait; ultra high resolution; vigilance; white matter

Protocol number 93-M-0170, NCT00001360 High resolution MRI In 2019, our Section published 5 papers that involved ultra-high resolution imaging. They were all carried out 7 Tesla (typical research is performed at 3 Tesla) which allows for higher resolution or smaller voxels because of the increased signal to noise ratio that comes with increased field strength. At 7T fMRI, spatial resolution has been pushed to the sub-millimeter domain, making it possible to resolve functional activity across cortical depths/layers. This approach faces technical constraints such as limited sensitivity, sensitivity to larger veins causing layer-specific microvasculature signal to be washed out by the dominant signal from ascending and pial veins, and misregistration between separately acquired anatomic reference images and functional images. To address these problems, we used a novel pulse sequence developed by Yuhui Chai, called VASO and PERfusion (VAPER) contrast acquired by combining the blood-suppression module of DANTE (Delay Alternating with Nutation for Tailored Excitation) pulse trains with 3D-EPI. This new sequence shows a highly specified functional layer profile and an improved sensitivity compared to current established laminar fMRI approach like VASO since it includes perfusion contrast along with volume contrast. To address the point of structural imaging, we introduced magnetization transfer (MT) weighted imaging which provides superior gray-white matter contrast so that all laminar fMRI analysis can be performed in the native EPI space, without the need for distortion correction and registration as the resulting anatomical data is completely matched for distortion with the VAPER. We demonstrated VAPER-fMRI using high resolution (0.8mm) to disentangle auditory and visual inputs in human planum temporale at submillimeter resolution over both dimensions of columnar distance and laminar depth. The integration of visual and auditory motion information is important to our ability to navigate the world. Planum temporale (PT) is part of the auditory cortex and it serves as a computational hub for processing multisensory information. We have found that PT area is topographically organized such that auditory and visual inputs consistently activate anterior and posterior subareas along the cortical ribbon respectively. Furthermore, along the laminar dimension, visual input led to response suppression to concurrent auditory input in anterior PT, most prominently in superficial layers. We also extended VASO imaging to allow for whole brain fMRI using sub-millimeter voxel dimensions, as previously, it was limited to single slabs in the brain. This allows us to address whole brain, layer specific connectivity questions. We openly shared our imaging sequence and analysis suite (open source) with anyone in the high-field community to be applied across various brain areas and the improved layer-specificity of our methods could be reproduced in more than 25 research labs worldwide. (https://layerfmri.page.link/VASO_worldwide). In the last year, we have continued our work on extending layer-specific fMRI methods to higher-order brain regions. Following our empirical paper published in 2019 in Nature Neuroscience, in which we demonstrated depth-dependent activity in human dorsolateral prefrontal cortex during a working memory task, we have developed a theoretical framework for adapting layer fMRI for the unique challenges and opportunities of association regions (whereas previously, applications had been limited to primary/unimodal motor and sensory areas). We have compiled best practices for data acquisition and analysis in these regions, as well as conceptual areas where layer fMRI can shed light on major outstanding questions in neuroscience (e.g., attention, consciousness, mental illness). These perspectives have been put forth in a manuscript that is pending at Progress in Neurobiology. Time Series Dynamics In the last year, we have developed a novel technique, called inter-subject representational similarity analysis (IS-RSA), to extract individual information from naturalistic fMRI data and shown that it can recover brain-behavior relationships while people watch complex, engaging videos. We have used this approach to demonstrate that individuals with higher fluid intelligence show more similar neural responses during movie-watching, while individuals with lower fluid intelligence display more variable patterns (i.e., less similar to one another and to high scorers). We have also used this approach to demonstrate that individuals with more similar responses to a personality questionnaire show more similar patterns of neural responses during movie-watching. Also along these lines, we have shown that functional connectivity calculated from data collected as subjects watch movies is more sensitive to trait-level behavioral differences than functional connectivity calculated during rest. In fully cross-validated models, both cognitive ability and affective traits could be predicted from as little as 2-3 minutes of functional connectivity data from movie-watching. Movies that were high in social content were particularly effective in generating accurate predictions. This result has implications for future large-scale data collection efforts aimed at brain-behavior predictive modeling, suggesting that using naturalistic paradigms in addition to or instead of resting-state acquisitions may hasten the development of translational tools based on fMRI functional connectivity. A manuscript on these results is in preparation. We have also designed, piloted, and begun data collection for a novel study that combines naturalistic tasks during fMRI with detailed behavioral and phenotypic assessment to test the hypothesis that patterns of brain activity during certain social videos selected based on extensive pilot behavioral studies to be particularly evocative but also ambiguous/open to different interpretations will stratify subjects into phenotypes relevant for depression and related mood disorders. Prior to the disruption of data collection due to COVID-19, we had acquired data from 40 healthy volunteers and 4 inpatients with severe depression in collaboration with Dr. Zarate's Section on the Neurobiology and Treatment of Mood Disorders. Analyses of these data are ongoing and we expect to resume data collection as soon as safety protocols allow. We have continued work to improve and better understand how multi-echo fMRI can be used to remove noise from fMRI data. Some of this work has focused on tedana.readthedocs.io software. In November 2019, SFIM organized and hosted a hackathon at NIH, which brought together the international collaborators of tedana for the first time. This work has greatly advanced this project and the communal accessibility and quality of multi-echo fMRI analysis methods. A brief summary of the accomplishments of the hackathon was shared with the tedana community at: https://tinyletter.com/tedana-devs/letters/tedana-hackathon-newsletter We have also been exploring methods for assessing ongoing vigilance state in the scanner to help reduce artifactual fluctuations in resting state fMRI. Recent work suggests that ultra-slow CSF fluctuations accompany descent into sleep. Here we evaluate how such fluctuations help track wakefulness in rest scans acquired on non sleep-deprived subjects using sequences not optimized for detecting such inflow-related fluctuations. We conclude that those fluctuations can be easily detected in other samples, and that they may provide valuable time-resolved information about fluctuations in wakefulness, as well as a means to segregate subjects according to their overall wakefulness levels.

协议号 93-M-0170，NCT00001360 高分辨率磁共振成像 2019年，我们组发表了5篇涉及超高分辨率成像的论文。它们均在 7 特斯拉（典型研究在 3 特斯拉）下进行，这允许更高分辨率或更小的体素，因为随着场强的增加，信噪比也随之增加。 在 7T fMRI 中，空间分辨率已提升至亚毫米级，从而可以解析跨皮质深度/层的功能活动。这种方法面临着技术限制，例如灵敏度有限、对较大静脉的敏感性导致层特异性微脉管信号被来自升静脉和软脑膜静脉的主要信号冲掉，以及单独采集的解剖参考图像和功能图像之间的重合失调。 为了解决这些问题，我们使用了柴宇辉开发的一种新型脉冲序列，称为 VASO 和 PERfusion (VAPER) 对比，通过将 DANTE（延迟交替章动定制激励）脉冲序列的血液抑制模块与 3D-EPI 相结合而获得。与目前建立的层流功能磁共振成像方法（如 VASO）相比，这种新序列显示出高度特定的功能层轮廓和更高的灵敏度，因为它包括灌注对比和体积对比。为了解决结构成像问题，我们引入了磁化转移（MT）加权成像，它提供了卓越的灰白质对比度，以便所有层流 fMRI 分析都可以在本地 EPI 空间中进行，而不需要像由此产生的解剖数据与 VAPER 的失真完全匹配。 我们展示了 VAPER-fMRI 使用高分辨率 (0.8mm) 在柱状距离和层流深度两个维度上以亚毫米分辨率解开人类颞平面的听觉和视觉输入。视觉和听觉运动信息的整合对于我们驾驭世界的能力非常重要。颞平面 (PT) 是听觉皮层的一部分，是处理多感官信息的计算中心。我们发现 PT 区域按地形组织，使得听觉和视觉输入一致地分别沿着皮质带激活前部和后部分区。此外，沿着层流维度，视觉输入导致前部 PT 对并发听觉输入的反应抑制，最明显的是浅层。 我们还扩展了 VASO 成像，以允许使用亚毫米体素尺寸进行全脑功能磁共振成像，与之前一样，它仅限于大脑中的单个板。这使我们能够解决整个大脑、特定层的连接问题。 我们与高场社区中的任何人公开共享我们的成像序列和分析套件（开源），以应用于各个大脑区域，并且我们方法的改进层特异性可以在全球超过 25 个研究实验室中重现。 （https://layerfmri.page.link/VASO_worldwide）。 去年，我们继续将特定层的功能磁共振成像方法扩展到更高阶的大脑区域。继我们于 2019 年在《自然神经科学》杂志上发表的实证论文（其中我们展示了工作记忆任务期间人类背外侧前额叶皮层的深度依赖性活动）之后，我们开发了一个理论框架，用于使层 fMRI 适应关联区域的独特挑战和机遇（而以前，应用仅限于初级/单峰运动和感觉领域）。我们编制了这些区域的数据采集和分析的最佳实践，以及层 fMRI 可以揭示神经科学中主要突出问题（例如注意力、意识、精神疾病）的概念领域。这些观点已在《神经生物学进展》待审的一份手稿中提出。 时间序列动力学 去年，我们开发了一种称为受试者间表征相似性分析 (IS-RSA) 的新技术，用于从自然功能磁共振成像数据中提取个体信息，并表明它可以在人们观看复杂、引人入胜的视频时恢复大脑行为关系。我们使用这种方法来证明，流体智力较高的个体在观看电影时表现出更相似的神经反应，而流体智力较低的个体则表现出更多的可变模式（即彼此之间以及与高分者之间的相似度较低）。我们还使用这种方法来证明，对性格问卷的反应越相似的个体在观看电影时表现出更相似的神经反应模式。 同样沿着这些思路，我们还表明，根据受试者观看电影时收集的数据计算出的功能连接比在休息时计算的功能连接对特质水平的行为差异更敏感。在完全交叉验证的模型中，认知能力和情感特征都可以通过观看电影时短短 2-3 分钟的功能连接数据来预测。社交内容丰富的电影在生成准确预测方面特别有效。这一结果对未来针对大脑行为预测建模的大规模数据收集工作具有影响，表明除了静息态采集之外或代替静息态采集，使用自然范式可能会加速基于功能磁共振成像功能连接的转化工具的开发。关于这些结果的手稿正在准备中。 我们还为一项新颖的研究设计、试点并开始数据收集，该研究将功能磁共振成像期间的自然任务与详细的行为和表型评估结合起来，以测试基于广泛的试点行为研究而选择的某些社交视频期间的大脑活动模式尤其如此的假设。令人回味但也模棱两可/对不同的解释持开放态度，将受试者分层为与抑郁症和相关情绪障碍相关的表型。在因 COVID-19 导致数据收集中断之前，我们与 Zarate 博士的神经生物学和情绪障碍治疗科合作，从 40 名健康志愿者和 4 名患有严重抑郁症的住院患者那里获取了数据。对这些数据的分析正在进行中，我们希望在安全协议允许的情况下尽快恢复数据收集。 我们不断努力改进并更好地了解如何使用多回波功能磁共振成像来消除功能磁共振成像数据中的噪声。其中一些工作重点关注 tedana.readthedocs.io 软件。 2019年11月，SFIM在NIH组织并主办了一场黑客马拉松，首次将tedana的国际合作者聚集在一起。这项工作极大地推进了该项目以及多回波功能磁共振成像分析方法的公共可访问性和质量。与 tedana 社区分享了黑客马拉松成就的简短摘要：https://tinyletter.com/tedana-devs/letters/tedana-hackathon-newsletter 我们还一直在探索评估扫描仪中持续警惕状态的方法，以帮助减少静息状态 fMRI 中的人为波动。最近的研究表明，超慢的脑脊液波动伴随着入睡。在这里，我们评估了这种波动如何帮助跟踪非睡眠剥夺受试者的休息扫描中的觉醒情况，使用未针对检测此类流入相关波动进行优化的序列。我们的结论是，这些波动可以在其他样本中轻松检测到，并且它们可以提供有关觉醒波动的有价值的时间分辨信息，以及根据受试者的整体觉醒水平来隔离受试者的方法。