Neonatal Brain Segmentation

描述(由申请人提供)：本申请涉及广泛的挑战领域(06)：使能技术，以及具体的挑战主题06-MH-103：神经科学研究的新技术。该项目的目标是开发全自动、准确的新生儿大脑分割方法，以促进对正常大脑发育的了解，以及未来对生命头两年神经发育障碍的研究。这一时期的脑发育是出生后脑发育最活跃的阶段，结构和功能发育迅速。尽管最近人们对儿童和青少年的大脑发育非常感兴趣，但由于这个年龄段儿童在脑图像获取和分析方面的挑战，人们对头两年的大脑发育知之甚少。因此，在理解早期大脑发育方面存在一个需要填补的知识空白。在北卡罗来纳大学教堂山分校，有一个专门为年轻儿科受试者成像的环境，包括一个由研究协调员、磁共振技术专家和护士组成的专门团队，北卡罗来纳大学的研究人员率先使用MRI对出生后头两年皮质和皮质下结构的灰质和白质成分的变化进行了纵向研究。特别是，由于NIH资助的四项研究侧重于早期大脑发育，已经获得了大量的纵向图像(包括两周时410张新生儿图像，一岁时185张图像，两岁时126张图像)。为了处理这种规模的数据，迫切需要需要最少人工干预的自动化工具，以更好地理解和研究生命头两年的早期大脑发育。然而，由于组织对比度低、图像质量差、组织强度随WM髓鞘变化而动态变化的特点，准确地从新生儿MR图像中分割出脑组织是非常具有挑战性的。此外，现有的成人甚至儿童脑图像分割算法不能很好地分割新生儿脑图像。我们建议通过开发一种新的新生儿分割方法来解决这一具有挑战性的问题，方法是利用我们四个由NIH资助的项目中获得的独特的纵向数据集。我们还提出了一种额外的基于图谱的新生儿脑分割方法，用于缺少纵向数据的新生儿图像，这在所有纵向研究中都是典型的。这些新生儿大脑分割方法的开发不仅将支持在北卡罗来纳大学进行的早期大脑发育研究，而且还将支持目前在其他研究所进行的其他类似项目，因为我们开发的算法将免费提供给研究社区，就像我们对锤子注册算法(http://www.nitrc.org/projects/hammer/)，)所做的那样，这是NITRC中下载最多的工具之一。通过这些开发的方法，我们最终将能够提取关于早期大脑发育的关键、以前未知的信息，这些信息将直接影响我们对正常大脑发育的理解，并将为未来研究和早期发现这一年龄段的神经发育障碍提供基础。 公共卫生相关性：该项目旨在通过开发、测试和评估两种全自动、准确的新生儿分割算法来解决新生儿磁共振脑图像分割这一具有挑战性的问题。这些工具的开发将有助于理解早期大脑发育和未来对生命头两年神经发育障碍的研究。此外，它还将有助于填补这一时期早期大脑发育的知识空白。

DESCRIPTION (provided by applicant): This application addresses broad Challenge Area (06): Enabling Technologies, and specific Challenge Topic 06-MH-103: New Technologies for Neuroscience Research. The goal of this project is to develop fully automatic, accurate neonatal brain segmentation methods, to facilitate understanding of normal brain development and also future study of neurodevelopmental disorders in the first two years of life. Brain development in this period is the most dynamic phase of postnatal brain development with rapid structural and functional growth. Although there has been a great deal of recent interest in childhood and adolescent brain development, very little is known about human brain development in the first two years, due to the challenges in brain image acquisition and analysis of children at this age. Therefore, there is a knowledge gap that needs to be filled in understanding early brain development. In UNC-Chapel Hill, with a special environment for imaging young pediatric subjects, including a dedicated team of study coordinators, MR technologists, and nurses, UNC investigators have pioneered using MRI to allow for longitudinal investigation of changes in the gray and white matter composition of cortical and subcortical structures in the first two years of life. Specially, with four NIH-funded studies focusing on early brain development, a large set of longitudinal images (including 410 neonatal images at two weeks, 185 images at one-year-old and 126 images at two-year-old) have been acquired. To process data of this scale, automatic tools that require minimal human intervention are highly needed for better understanding and study of early brain development in the first two years of life. However, due to low tissue contrast, poor image quality, and dynamic changes of tissue intensity with WM myelination, it is very challenging to accurately segment brain tissues from neonatal MR images. Moreover, the existing segmentation algorithms developed for adult or even pediatric brain images fail to segment neonatal brain images satisfactorily. We propose to address this challenging issue by developing a novel neonatal segmentation method by taking advantage of the unique longitudinal datasets acquired in our four NIH-funded projects. We also propose to develop additional atlas-based neonatal brain segmentation method for the neonatal images with missing longitudinal data, which is typical in all longitudinal studies. The development of these neonatal brain segmentation methods will support not only the early brain development studies conducted in UNC, but also other similar projects currently conducted in other institutes, since our developed algorithms will be made freely available to the research community, as we did with our HAMMER registration algorithm (http://www.nitrc.org/projects/hammer/), which is one of the top downloaded tools in NITRC. With these developed methods, we will eventually be able to draw critical, previously unknown information about early brain development, which will have direct impact on our understanding of normal brain development and will provide the basis for future study and early detection of neurodevelopmental disorders in this age group. PUBLIC HEALTH RELEVANCE:This project aims at addressing a challenging problem of neonatal MR brain image segmentation by developing, testing, and evaluating two fully automatic, accurate neonatal segmentation algorithms. The development of these tools will help the understanding of early brain development and future studies of neurodevelopmental disorders in the first two years of life. Furthermore, it will help fill the knowledge gap of early brain development in this period.