PNAs: development of white matter pathways in human brain during the second and third trimester of pregnancy

Abstract ： In the middle and late stages of human pregnancy , Rapid neural development consists of neuronal migration 、 Cellular tissue 、 Supported by basic processes such as cortical stratification and myelination . At this time , The growth and maturation of white matter lay the foundation for an efficient structural connection network . There is limited detailed knowledge about the brain development trajectory of healthy human fetuses , Part of the reason is to get high-quality MRI Data has inherent challenges . ad locum , We use state-of-the-art high-resolution multi shell motion correction diffusion weighting MRI(dMRI), As a developing human connectome project (dHCP) Part of , To represent 113 individual 22 - 37 The maturation of the white matter microstructure of the fetus in the uterus during the first week of pregnancy . We defined 5 Major white matter tracts , The traditional diffusion tensor model and multi shell and multi structure model were used to characterize the microstructure characteristics . Compared with correlation bundles , We found a unique maturation trend in the thalamic cortical fibers , Different maturation trends were found in specific parts of the corpus callosum . Although the partial anisotropy of the pressing part of the corpus callosum increases linearly , However, the partial anisotropy of most other white matter bundles shows a complex nonlinear trend , In the first trimester of pregnancy, partial anisotropy first decreases , And then added . The latter deserves special attention , Because it is significantly different from the trend previously described in extrauterine preterm infants , This suggests that such normal fetal data can provide important insights into the connectivity abnormalities of neurodevelopmental injuries associated with preterm birth . 1. sketch In the human fetus , The development of the main white matter pathways in the brain is extremely rapid in the second and third stages before pregnancy and has a clear hierarchical order . The structure and integrity of these white matter connections play an indispensable role in supporting and coordinating functional networks . The current understanding of these processes relies heavily on posthumous data . The fetus MRI It can capture the development of the whole brain in its living and functional state , This provides important additional information for understanding normal growth . Especially white matter , This may include the development of remote connections and a detailed survey of the trajectories of specific areas . The high prevalence of cognitive and motor problems in preterm infants underscores the importance of better understanding this critical period . Among these babies , Early exposure to the extrauterine environment may affect later neurodevelopmental trajectories . Many evidences show that , White matter abnormality is the main pathology , To show further that , This particular type of organization is at a critical stage of development , And vulnerable to external influences . under these circumstances , Characterization of white matter maturation in uterus has an important normative reference role . Because it is difficult to obtain inherent imaging data from this population , For example, processing image artifacts related to maternal tissue and continuous fetal movement , Therefore use MRI It is challenging to accurately characterize the development of fetal white matter tracts in vivo by other non-invasive methods . It is also necessary to recruit enough subjects to explain the population heterogeneity and age effect . Previous studies are difficult to generalize as the representative of typical development , Because they include clinical populations of brain abnormalities or extrauterine preterm infants . All existing studies use diffusion tensor imaging (DTI) To describe changes in microstructure ; However , The results are not consistent . Although some studies have reported DTI Index and gestational age (GA) The linear relationship between , Other studies fit the nonlinear model , Other studies have not found a significant age dependence . In this study , We use the most advanced high angle resolution multi shell diffusion weighting MRI (dMRI) Acquisition technology to solve DTI Limitations and challenges of fetal imaging , And the reconstruction and processing pipeline specially developed for the study of challenging fetal data , This is the developing human connectome project (dHCP) Part of (http://www.developingconnectome. org). We apply newly developed and optimized methods , In a large queue from 22 To 37 Weekly 113 A healthy fetus . By these means , We can depict specific white matter tracts , Including left and right corticospinal tracts (CST)( An example of a projected beam ), Visual radiation beam (ORs) And the lower longitudinal fasciculus (ILF)( Examples of joint bundles ), And the corpus callosum (CC)( Examples of commissural bundles ). These specific bundles were chosen because of known differences in their developmental trajectories , And their injury or abnormal development is related to the pathophysiology of neurodevelopmental disorder or intellectual disability . This study is the largest and most detailed intrauterine study on the mature changes of white matter microstructure in the middle and late stages of human pregnancy , It provides us with valuable resources to improve our understanding of the neurophysiological basis of neurodevelopmental disorders . 2. result 2.1 Whole brain development and... In the fetal cohort FA Specification trend of The fetus dMRI The data were collected in 151 example (22 - to 38 Zhou ) Of dHCP. Each subject used dHCP Preprocessing pipeline for processing , This includes considering the unpredictability of fetal movements 、 Geometric distortion of echo plane imaging 、 Specific measures such as uneven signal intensity caused by fetal position difference . The head of the fetus is small , Far away from the coil , Poor signal-to-noise ratio . In the process of manual evaluation 151 Of the subjects ,38 Subjects failed due to excessive exercise during the collection process . To verify that the data set shows the expected trend of normal capacity growth , We calculated the relationship between total brain volume and gestational age for each subject . Consistent with existing literature , We found that throughout the study , The volume increases linearly . The whole brain average FA And GA In a similar linear relationship ( chart 1C).

chart 1 （A） Fetal age distribution ;（B） Whole brain volume plotted according to gestational age ;（C） The whole brain average plotted according to gestational age FA. 2.2 22-37 Weeks gestational age projection beam 、 Combined and commissural beam microstructures Next , We are MRtrix3 Constrained spherical deconvolution is used to estimate the density function in a single direction (odf). This method solves the challenge of solving the cross fiber group in one voxel , Improved bundle graph estimation . This may confuse other commonly used methods , Such as diffusion tensor imaging . First of all, a single subject's odf Collected in the average template for each gestational week , Then each 3 Zhou (22、26、29、32 and 35 Gestational weeks ) Use probabilistic streamline beam tracing to sketch 5 There are different white matter pathways (CC、CSTs、ILFs、ORs). Except in the smallest 22 Difficult to estimate in the fetal circumference template OR Outside , Fiber tracing was successful in all cases , But it can be reliably identified at all other ages ( chart 2). Then the twisted transformation from template to subject is used to convert the fiber bundle from age-matched template to individual subject space .

chart 2 White matter pathway obtained by target probability streamline tracing 2.3 White matter bundles have different maturation trajectories In order to put our findings in the context of previous studies on white matter development , We first use the dispersion tensor measure ; FA And average diffusion coefficient (MD) To estimate the change of microstructure under each fiber bundle . Average FA and MD In different white matter bundles, there are different maturation trajectories ( chart 3).GA In most of the circumscribed white matter tracts, the relationship between and tensor measures is best described by quadratic polynomial fitting ( chart 3), The pressure part FA、MD And CST in MD Linear relationship . As expected , Throughout the study period , There are complex developmental changes in white matter ,FA/MD and GA The relationship between them is significant in all bundles . Left 、 There was no significant difference between the right hemispheres in any of the defined brain tracts . Of particular note , stay CC The obvious FA/MD Maturity trends , At the pressing part FA and GA There is a linear relationship between (ρ =0.36), But the relationship at the knee is more complicated .FA Value from 22 To 30 Zhou GA First lower , Then it increases after full-term pregnancy ( chart 3). stay MD This relation in the value is the reciprocal relation , The pressure part decreases linearly (ρ= 0.3) It has a similar nonlinear relationship with the knee , The peak is 30 Zhou , Then decrease .CC Pressure part ,FA from 22 To 30 The weekly trend is downward , then CST,ILF,OR from 30 Full term is a stable slope ( chart 3). Again , stay MD and GA The relationship between them is also a reverse trend in these fiber bundles . from 22 Thoughtful 30 Zhou , It starts to rise , Then it drops at full term . The exception is CST, It shows a strong negative correlation (ρ = 0.75). For completeness , We also calculate the axial and radial diffusion coefficients for each region , These figures can be found in the supplementary information section .

chart 3 about CST,ILF,OR, Pressure part , Average knee FA and MD 2.4 The tensor measure is proved by the multi shell and multi organization model Considering the relatively small size of the fetal brain , Part of the volume of the tissue may affect the FA and MD value , Especially when a streamline crosses a voxel , These voxels contain both white matter 、 Gray Matter , It also contains cerebrospinal fluid (CSF). To specifically address these partial volume effects , And see if they are responsible for the maturity trends we observe , We are right. DWI A spherical deconvolution model with multi shell and multi tissue constraints is applied to the data , The unique use of white matter and cerebrospinal fluid signals b Value dependence is used to describe the endogenous contribution of brain tissue and fluid . As expected , The analysis found that , In all the enclosed bundles , Fluid content and MD There is a strong positive linear trend between ( chart 4), Average FA There is a positive correlation with tissue anisotropy ( chart 4). It is important to , These linear trends indicate , The nonlinear maturation trend observed in our data cannot be attributed to simple partial volume effects . In order to highlight GA in MD And fluid fraction trends , Can be in SI appendix ( chart S2) Find the fluid fraction and GA The relationship between .

chart 4 Average fiber bundle MD Partial correlation with volume fraction of liquid components （ On ）, Average of fiber bundles FA（ Next ） 3. Discuss In this work , We use the uterus dMRI Reported 113 individual 22 - 37 Development of white matter in fetal brain during gestation , As open access dHCP Part of . This represents the largest cohort of fetuses studied in this way , We use the most advanced collection and analysis methods , So far, it is possible to describe the white matter pathway in fetal brain in the most detail . Use these methods , We studied the main connections of the brain 、 Projecting and crossing fibers , It is proved that each species has different developmental ontogeny , Some of them showed nonlinear changes in the beam microstructure throughout the study .