【作者】 展金锋；

【导师】 刘树伟； Arthur W.Toga；

【作者基本信息】 山东大学 ， 人体解剖与组织胚胎学， 2013， 博士

【摘要】 胎儿大脑在结构和连接上与新生儿及成人大脑都有很大的不同。随着磁共振成像(MRI)技术的进步,胎儿磁共振在临床产前检查上应用的越来越广泛,并且逐渐成为研究大脑发育和成熟的重要手段。MRI具有同时显示胎儿大脑解剖和组织结构的优势,可对胎儿脑层状结构和脑沟精确成像。因此探讨与大脑发育相关的MRI信号改变十分重要,尤其是针对一过性结构的出现和消失、潜在的组织学改变以及发育过程中时间标志的研究。越来越多的形态学后处理方法和软件工具使得胎儿MRI研究成为可能,尤其是模板的应用显著提高了大脑MRI数据自动分析的准确性和效率性。但是胎儿大脑在大小、形态和结构上的发育很快,甚至每周都有改变,单个模板很难满足不同孕周的需要。现有的大多数研究的样本量局限,年龄跨度较窄,因此建立时间-空间模板对于分析早期大脑的动态发育就变得十分有必要。妊娠第二期从14孕周到27孕周,是大脑结构变化最多最快的时期,该阶段也被称为胎儿脑发育的敏感期。在这个过程中,大量的神经发生和神经迁移快速进行,即使相对小的干扰也会导致大脑成熟后严重的结构和功能改变,因此针对这一时期大脑正常和病理性发育的研究对于认识病因学以及神经病学特征和不同环境因素之间的关系至关重要。在临床上,胎儿磁共振通常应用在19孕周以后,因此现有的大多数胎儿磁共振研究,如时间-空间模板的建立、三维重建、组织分割等,都集中在妊娠第二期的后期及之后的阶段,尚缺乏涵盖第二妊娠早期的胎儿大脑模板。活体胎儿磁共振存在很多局限性,如胎儿脑小、扫描序列的选择、胎动等,很难获得显示解剖细节的高质量图像。但是标本研究可以给我们提供很多优势,如场强大、视野小、层厚薄、扫描时间长等,并且很多研究已经表明标本的组织学结构与MRI信号标记的一致性。本研究旨在利用磁共振在胎儿标本大脑成像方面的优势,为大脑早期发育的形态学改变提供定量研究的模板,并利用该模板探索妊娠第二期早期大脑成熟过程中各形态结构的发育类型及机制,并为早期胎儿脑发育的临床诊断提供参考。第一部分：15-22孕周胎儿脑模板的建立目的：时间-空间模板的应用可以显著提高胎儿磁共振数据的分析结果和效率,本研究的目的在于建立妊娠二期早期(15-22孕周)的年龄特异性的胎儿大脑模板。材料与方法：收集34例15-22孕周的胎儿标本,进行7.0T高场强磁共振扫描,利用加州大学洛杉矶分校神经影像实验室(LONI)的Pipeline流程软件进行格式转换和偏移场校正,使用BrainSuite软件手动去除非脑组织,使用Pipeline的形态分析软件包计算每个胎儿脑的表面积、体积、形状指数和曲度,使用宾夕法尼亚大学的先进的标准化工具(ANTS)中的模板建立算法建立每个孕周的平均脑模板,最后基于每个周的模板建立最终模板。结果：获得从15到22孕周的胎儿大脑形态学测量值,其中大脑体积和表面积的发育呈线性回归,体积增长约4倍,表面积增长约2.5倍。为了减小每个孕周的胎儿数量不同而导致的权重偏倚,我们首先建立每个周的平均模板,使用这8个孕周的模板,15-22孕周的平均模板被建立,胎儿脑层状结构可以观察到4层,从外向内分别是皮质层、皮质下区、中间区和脑室区；基底节的各个组成部分也可以分辨。结论：本研究首次建立了15-22孕周的时间-空间胎儿脑模板,并获得了大脑早期发育的解剖结构变化曲线。高场强磁共振数据的使用使得该模板具有较高分辨率和对比度,将有助于对胎儿脑解剖发育的动态改变的进行形态学分析。第二部分：胎儿脑层状结构和皮质下结构的发育目的：层状结构是胎儿大脑重要的解剖特征之一,本研究的目的在于利用在模板建立过程中的产生的形变值,分析层状结构及皮质下结构在15-22孕周的发育改变。材料和方法：将每一个胎儿脑配准到第一部分建立的模板中,使用ANTS将配准过程中生成的仿射和弯曲变形值合并成一个变形场,然后转换成Jacobian值,使用植入LONI Pipeline的FSLstats统计Jacobian值,变形场的区域结构改变通过基于张量的形态学方法(TBM)展示,不同脑区的局部定量改变与孕龄做线性回归分析,使用植入Pipeline的在线统计计算资源(SOCR)分析基于张量的形态发育统计图。结果：除了在额叶和顶叶的部分脑室区和中间区,几乎所有的大脑皮层都呈现出显著的增长。其中,皮质下区的发育增长与孕龄有更明显的相关性和更大的生长速率,同时表现出区域发育的不均质性。但脑室区,包括神经节凸起(GE),没有显著的改变。额叶和顶叶区的侧脑室表现出负相关和负生长速率,表明侧脑室的体积相对缩小。脑干发育与孕龄的相关性比基底节高,但基底节的生长速率更大。结论：本研究首次基于群体数据统计分析了该阶段胎儿脑层状结构的发育规律,结果表明皮质下区在所有层状结构中表现出最明显的生长,并且是这一阶段整个层状结构发育的主导因素。第三部分：基于优先信息和模板的层状结构分割目的：通过结合手动和自动的分割方法,基于优先信息和模板,对胎儿脑的层状结构进行分割,建立层状结构的组织概率性时间-空间模板。材料和方法：使用ITK-SNAP对15-22孕周总模板的层状结构进行手动分割,将层状结构分为四层,从外向内分别是皮质层、皮质下区、中间区和脑室区。手动分割后的标记图通过ANTS的逆向变形过程转换成每一个模板的层状结构分割标记。然后基于所获得的标记图作为优先信息,使用ANTS的Atropos对每个孕周的模板进行自动分割,最后使用ITK-SNAP计算四层结构的体积,并使用线性回归模型统计分析各层的组织强度随孕龄的变化。结果：建立了层状结构的概率性图谱,精确描绘了皮质层、皮质下区、中间区和脑室区这四层结构,对每一层的3D网格表面重建使我们可以直观的观察各层的发育过程,通过定量分析发现每一层的体积都随孕龄增加而增加,其中皮质下区增长最显著,各层的图像强度的平均值和标准差随孕龄下降,表明每一层的组织学组成变得更加一致。结论：使用手动分割结果作为优先信息可以提高基于模板组织分割的精确性,帮助获取各层状结构生长相关的改变,其中皮质下层的增长最为显著,整个层状结构趋于成熟化。第四部分：外侧裂的发育及胎儿脑的发育方向目的：描绘发育最早的脑沟-外侧裂的早期发育规律,并寻找脑沟形成的内在机制；另外,通过描绘发育过程中皮质表面的形态改变,探讨大脑整体和不同脑叶的发育规律。材料和方法：使用同样的数据和第一部分生成的模板及第二部分基于张量的统计图,通过快速进程算法计算分配距离函数,进而将以三角面片为代表的表面转化成内嵌表面以代替常规的表面网格化,最后在欧氏空间内计算整个脑表面的平均曲度,以观察外侧沟的折叠过程。另外,通过微分同胚算法将每一个样本的皮质表面配准到模板表面,然后使用局部形态分析的Pipeline工作流程计算局部形态测量值,获得每一个三角化表面顶点的位移场和放射距离,最后使用在线统计计算资源(SOCR)将每一个皮质点上的位移场和放射距离值作为协变量与孕龄做回归分析,从而获得发育过程中皮质表面在局部的膨胀或收缩信息。结果：获得大脑表面折叠度的定量发育改变,外侧沟早期呈开放、浅而钝的外形,之后变得相对封闭、深而长,折叠度不断增加,外侧沟周围的皮质区,尤其是额叶和顶叶区的相应皮质,与孕龄有更显著的相关性和更大的生长速率。对于位移场,颞叶前端的改变最为显著,整个皮质表面前后表面以正相关为主,上下表面以负相关为主。对于放射距离,整个大脑表面为正相关,但额叶和颞叶的前端以及枕叶后端具有更显著的相关和更大的生长速率。结论：外侧沟周围皮质的快速生长参与了外侧沟的折叠形成过程。从15到22周,大脑表面不断向外膨胀,但这一阶段主要以前后方向的扩张为主,即以额叶、颞叶和顶叶的生长发育为主。更多还原

【Abstract】 The fetal brain is substantially different from the neonatal brain in terms of its structure and connectivity. Driven by the ongoing magnetic resonance imaging (MRI) techniques progress, fetal MR imaging has been more and more applied in clinical prenatal diagnosis. Moreover, it also become an essential tool to study fetal brain development and maturation. MRI can provide information both about gross anatomical structures as well as histological microstructures, lamination and sulcation for example. It is essential to understand MR signal changesassociated with maturation, including the appearance and disappearance of transient structures, the underlying histological developmentof thefetal brain as well as the timing of developmentof landmarks in maturation. The availability of post-acquisition morphometric methods and powerful new software tools enable the study of early fetal brain development and maturation. The use of atlases can significantly improve the accuracy and efficiency of automated analysis of brain MRI data. However, fetal brain undergoes more changes in size, shape and structure than at any other time in life, even in every week. A single atlas can’t meet the requirement of different gestational ages. Most of studies have a limited number of subjects, a narrow age range. Therefore, building a spatio-temporal atlas to model the dynamic changes during early brain development becomes very necessary.The second trimester ranges from14th week of pregnancy to27th week, and is considered to be a specific window of vulnerability for the fetus. During this period, the enormous neurogenesis and neuronal migration are happening, so relatively minor disruptions may significantly alter the structure and function of the maturing brain. Studies of normal and pathological brain development during this period are critical for our understanding of the etiology and the associations between neurological traits and different environmental phenotypes. In clinical settings, fetal MRI is usually performed after the19th gestational week. Therefore, most fetal MRI studies, construction of spatial-temporal brain atlases,3D reconstruction, tissue segmentation etc. for example, are focused on the later period of the second trimester and afterwards. Fetal atlases covering age-range of the early second trimester of pregnancy are currently missing in the scientific community.Fetal MRI in vivo has many restrictions due to the smaller brain-size, sequence selection, and frequent fetal generic movements. It is difficult to obtain high-quality images with detailed local anatomy. However, postmortem fetal specimen offer advantages by allowing the use of high-field strength magnets, smaller field of view, reduced slice thickness, and increased acquisition time. Some pioneering studies have demonstrated the correspondences between histological structures and MRI bioimaging markers. This study aims to provide fetal brain atlases for quantitative assessment of morphometric brain changes, yield clues to the underlying brain maturation patterns and mechanisms during the early second trimester, and provide references for clinical diagnosis during the early fetal brain development.Part1:Establishment of fetal brain atlases during the early second trimesterObjective:The use of spatiotemporal atlases can significantly improve the results and efficiency of automated analysis of fetal brain MRI data. The current study aims to establish age-specific fetal brain atlases of the early second trimester (15-22gestational weeks)Materials and Methods:The thirty-four postmortem fetal specimens of15-22weeks GW were collected and performed in a7.0T Micro-MRI. Both the data conversion and bias field correction were implemented as Pipeline workflows developed by Laboratory of Neuroimaging of UCLA. Manual removal of the non-brain tissue was performed using BrainSuite software. For each cortex,4complementary global shape metrics were computed using LONI ShapeTools pipeline library-volume, surface area, shape index and curvedness. The script buildtemplateparallel.sh of Advanced Normalization Tools (ANTs) developed by University of Pennsylvania was run in each week to build the average template of each week. Three-Dimension surface reconstruction of these templates were performed by BrainSuite. The general template was built from these weekly templates.Results:From15to22gestational weeks, growth trajectories of brain area and volume is fitted well by linear regression model. The whole brain increased in volume by approximately4-fold and about2.5-fold in area. To reduce the bias effects of different demographic distributions of different numbers within each week, we built the templates each week first. After3D surface reconstruction of these templates, we found that the15th week still has some trails of neural folding in the early stages. The brain surface of22nd week still looks smooth, but the whole brain’s general shape looks more mature. The aggregated general template was then built based on these weekly templates. Four layers of lamination structures are displayed, from outer to inner, these layers are: cortical plate, subplate zone, intermediate zone, ventricular zone. The composed parts of basal ganglia could also be distinguished.Conclusion:Our study first established the spatial-temporal atlas of the early second trimester. And we also obtained the general developmental trajectories of the early fetal brain development. These atlases based on the high-field MRI data provide good resolution and contrast, which will enable to characterize the dynamic anatomical changes of fetal brain development. Part2:Development of lamination structure and subcortexObjective:Lamination structure is one of the most important characters of fetal brain. This study aims to make the use of transform values during the atlas building to analyze the age-specific changes in the patterns of lamination and subcortex structures from15to22gestational weeks.Materials and methods:Each subject’s brain was registered to the corresponding template built at the first research part. Using ANTS, the affine and warp transform files of each subject computed during the registration were composed to one deformation file, and then calculated to be Jacobian. Using FSLStats module in LONI Pipeline, we obtained the statistics of Jacobian field.Regional structural differences of deformation fields were represented as (Tensor-based morphometry) TBM maps. The general linear model was used to study the associations between localized quantitative deformation of different brain structures and gestational weeks. The computational libraries of the Statistics Online Computational Resources (SOCR) wrapped as Pipeline modules were used for computing shape-based statistics.Results:Nearly the entire lamination grows significantly, except for parts of ventricular zone and intermediate zone in the frontal and occipital lobes. Compared with other layers, the subplate zone show more significant correlation with GW and a bigger growth rate. Regional developmental differences could be found within the subplate zone. However, the ventricular zone (including ganglionic eminence) does not show significant changes. Lateral ventricles in the frontal and parietal lobes show negative correlation and growth rate, which implies local volume contractions. Thalamus with a higher correlation than basal ganglia. But basal ganglia has a bigger growth rate.Conclusion:Based on population-base statistics, this study first analyze the development trajectories of lamination structures during the early second trimester. The results showed the subplate zone shows most obvious growth pattern, and is the predominant cause of the changes in the lamination pattern of the cerebral wall. Part3:Prior and atlas-based segmentation of the laminar organizationObjective:This study aims to segment the laminar organization bycombining manual and automatic methods with prior and atlas-based segmentation, to construct spatiotemporal atlases of the fetal brain with temporal models of tissue probability of laminar layers.Materials and methods:ITK-SNAP tool was chosen to do the manual segmentation. Laminar organization of cerebral wall in the overall atlas, which represents the structures of fetal brain between15and22GW, was segmented for four layers.From pial surface to ventricular, they are: cortical plate (CP), subplate zone (SP), intermediate zone (IZ), ventricular zone (VZ)-After manually segmented label image was finished at the overall atlas, segmented image was transformed to the segmentation information of each week’s atlas. This process is realized by use of opposite direction of WarpImageMultiTransform in ANTS. Segmentation is realized byAtroposin the ANTS, based on the prior information of inversewarped segmentation.We then calculated volumes of the segmented structures, as well as the statistics of the image intensity for each structure using ITK-SNAResults:The probabilistic maps of laminar layers are created and we create a set of accurate delineations of four layers including cortical plate, subplate zone, intermediate zone and ventricular zone.3D mesh surface and quantitative measurements of each layer are displayed and analyzed, which allow us to capture the appearance, disappearance and spatial variation of brain structures over time. Thevolumes of every layer grow steadily with increasing ages, and subplate zone shows a more significant growth characteristic.The mean intensity and standard deviation decrease with GW imply that the composition of each layer is becoming more and more uniform.Conclusion:Experimental results indicate that subplate zone shows most obvious growth pattern and using manual segmentation as prior can correctly capture growth-related changes in the fetal cerebral wall and provide improvement in accuracy of atlas-based tissue segmentation.Laminar organization grows toward mature during the period we study. Part4:Sylvian fissure development and growth directionObjective:This study aims to characterize the developmental trajectory of Sylvian fissure, the earliest sulcus during the early brain development. And we also explored the mechanisms of sulcal formation during development. Besides, by analyzing changes of cortical surface deformation during development, we explored the global and different lobes growth pattern.Materials and methods:We used the fast marching algorithm to compute the signed distance function, and then converted the triangular mesh representation to implicit representations of surfaces in replacement of regulargrid. Then we computed the mean curvature on the surface on Euclidean space to observe the folding process of Sylvian fissure. We performed surface registration of the template cortical model to each individual cortical surface using a diffeomorphic algorithm. Then local shape analysis (LSA) pipeline workflow was used to obtain local shape metrics (displacement-field, radial-distance), per vertex in the triangulated surface representations. Using SOCR, these two local shape measures were used as covariates with gestational week to generate brain maps at each cortical location, and captured local expansions or contractions of the developing cortical surface.Results:Development process of brain surface was demonstrated by mean curvature. And the Sylvian fissure is widely open, obtuse and shallow in the early stage. Subsequently, it becomes narrow, deeper and longer, and its mean curvature increased continuously. The cortex around the Sylvian fissure, especially on the frontal and parietal lobes, has a higher growth rate and more significant correlation. The most significant changes in displacement-field were observed in the anterior region of temporal lobe. For the entire cortical surface, there are more vertices with positive correlation values in both the anterior and posterior areas, but more negative ones superiorly and inferiorly. For the radial-distance metric, there are only positive correlation values. However, vertices in the anterior surface of the frontal lobe, the temporal lobes and the posterior surface of the occipital lobe have higher growth rate and more significant correlation.Conclusion:The faster growth of the cortex contributes to the folding process of the Sylvian fissure. Every vertex on the surface undergoes radially expanding from15to22GW. The brain shape extends in both the anterior and posterior directions during this period, In particular the development of the frontal, temporal and occipital lobes.更多还原