CRCNS Research Proposal: Coupled Learning for Anatomically and Developmentally Consistent Analysis of Macaque-Human Fetal Brain Growth

神经发育障碍，如自闭症谱系障碍、注意力缺陷/多动障碍、胎儿酒精谱系障碍以及与早产相关的并发症，会在整个一生中影响受影响个人的生活质量。核磁共振成像(MRI)已经确定了受这些疾病影响的人和“典型发育”个人之间的神经解剖学差异，但导致这种差异的生物学机制及其与疾病过程的联系尚不完全清楚。对孕妇进行磁共振成像是安全的，而且在图像采集过程中解释胎儿运动的能力最近的发展，使得胎儿大脑的高分辨率3D成像成为可能。为了更好地理解MRI观察到的人类解剖变化轨迹背后的发育机制，还收集了非人类灵长类动物的纵向测量数据。人类和非人类灵长类动物在大脑结构和功能上有许多相似之处，这使得在一种动物身上的发现可以转化为另一种动物。非人灵长类动物研究的优势是，许多因素在人类怀孕期间可能会有所不同，可以通过实验进行控制，并且可以进行更详细的纵向成像。这项研究将开发计算方法，更精确地将人类和非人类大脑之间的胎儿发育联系起来。这项工作利用独特的人类和非人类灵长类成像数据集，通过新的方法系统地将大脑标记为相应的子区域，并在发育事件之间建立更紧密的联系。这种提高的精确度将增强从正在进行的人类观察研究中获得的知识，并使解决神经发育疾病的新临床方法成为可能。利用磁共振成像(MRI)非侵入性地监测人类和非人类灵长类动物的胎儿大脑发育的能力，为通过纵向实验设计表征大脑发育提供了新的机会。然而，考虑到获取数据的频率增加和高分辨率图像的质量，一个重要的新限制是无法以所获取的数据的精确度来转换物种之间的发育时间点。用于研究出生后脑图像的传统方法利用处理步骤，例如对公共解剖坐标系的空间归一化，分割成组织类别，以及分割成已知的神经解剖区域。要使这些技术适用于研究发育中的胎儿大脑，需要对数量进行年龄和物种特定的定义，如瞬时发育区，或皮质回和脑沟的出现。该项目利用日益强大的机器学习技术，并利用目前正在收集的日益丰富的胎儿成像数据，来提取一致的跨物种大脑发育测量。另一个目标是制定精细的、解剖学上和时间上跨物种一致的定义。这些神经解剖学上的局部定义将被用来量化这两个物种在胎儿大脑发育过程中的局部形态生长。这项工作为支持胎儿大脑发育的神经成像研究的计算科学和神经科学做出了贡献。这些进展将提供一种新的翻译资源，将有关大脑发育的解剖学和时间上的特定信息联系起来，包括正常生长和专注于神经发育障碍的临床和动物模型实验。该奖项由CEISE信息和智能系统(IIS)通过CRCNA和大脑计划以及MPS数学科学部(DMS)通过数学生物学计划共同资助。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Neurodevelopmental disorders such as autism spectrum disorder, attention deficit/hyperactivity disorder, fetal alcohol spectrum disorders, and complications associated with premature birth, impact the quality of life of affected individuals over the entire lifespan. Neuroanatomical anatomical differences between people affected by these conditions and “typically developing” individuals have been identified with magnetic resonance imaging (MRI), but the biological mechanisms leading to such differences, and their link to the disease processes, are incompletely understood. It is safe to perform MRI on pregnant women, and recent developments in the ability to account for fetal motion during image acquisition, have enabled high-resolution 3D imaging of the fetal brain. In order to better understand the developmental mechanisms that underlie the trajectory of anatomical changes observed by MRI in humans, longitudinal measurements are also collected in nonhuman primates. Human and nonhuman primates share many similarities in both brain structure and function that allow findings in one to be translated to the other. Advantages of nonhuman primate studies are that many factors that may vary between human pregnancies can be experimentally controlled, and much more detailed longitudinal imaging is possible. This research will develop computational approaches to more precisely link fetal growth between human and nonhuman brains. The work leverages unique human and nonhuman primate imaging datasets with new methods for systematically labeling the brain into corresponding sub-regions and establish closer links between developmental events. This increased precision will enhance knowledge gained from ongoing human observational studies and enable new clinical approaches to address neurodevelopmental diseases. The ability to non-invasively monitor fetal brain growth in both human and nonhuman primates using magnetic resonance imaging (MRI) provides a new opportunity to characterize brain development with longitudinal experimental designs. However, given the increased frequency with which data can be acquired, and quality of high-resolution images, an important new limitation is the inability to translate developmental time points between species at the level of precision of the acquired data. Conventional approaches for studying postnatal brain images utilize processing steps such as spatial normalization to a common anatomical coordinate frame, segmentation into tissue classes, and parcellation into known neuroanatomical regions. Adaptation of these techniques to study the developing fetal brain requires age and species-specific definitions for quantities such as transient developmental zones, or emergence of cortical gyri and sulci. This project makes use of increasingly powerful machine learning techniques and leverages the increasingly rich fetal imaging data now being collected, to extract consistent cross-species measures of brain development. An additional objective is to develop fine scale anatomically and temporally consistent definitions across species. These neuroanatomically localized definitions will then be used to quantify regional morphometric growth during fetal brain development in both species. This work contributes to the computational science and the neuroscience that supports neuroimaging studies of fetal brain development. These developments will provide a new translational resource to link anatomically and temporally specific information about brain development, both in normal growth and in clinical and animal model experiments focused on neurodevelopmental disorders.This award is being co-funded by the CISE Information and Intelligent Systems (IIS) through the CRCNA and BRAIN Programs, and the MPS Division of Mathematical Sciences (DMS) through the Mathematical Biology Program.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.