Mapping brain maturation

Abstract

Human brain maturation is a complex, lifelong process that can now be examined in detail using neuroimaging techniques. Ongoing projects scan subjects longitudinally with structural magnetic resonance imaging (MRI), enabling the time-course and anatomical sequence of development to be reconstructed. Here, we review recent progress on imaging studies of development. We focus on cortical and subcortical changes observed in healthy children, and contrast them with abnormal developmental changes in early-onset schizophrenia, fetal alcohol syndrome, attention-deficit-hyperactivity disorder (ADHD) and Williams syndrome. We relate these structural changes to the cellular processes that underlie them, and to cognitive and behavioral changes occurring throughout childhood and adolescence.

Figure 1

Typical processing steps in an analysis of MRI brain scans. (a) A typical coronal section from a T1-weighted MRI scan of the brain. (b) The result of applying a tissue-classification approach to classify image voxels as gray matter (green), white matter (blue) or cerebrospinal fluid (CSF; red). Non-brain tissues such as scalp and meninges surrounding the brain have been digitally edited from the image. (c) Parcellation of the brain into the frontal lobe (blue), parietal lobe (green), occipital lobe (red) and temporal lobe (yellow). This subdivision of anatomy is performed with the aid of a cortical surface model on which sulcal landmarks separating the lobes can be reliably identified. Once partitioned in this way, the volumes of each tissue type in the major lobes can be computed and growth curves established for each major lobe. The quantity of gray and white matter in the brain and of CSF in the ventricles and cortical sulci can be computed and compared across subjects and over time. Maps of these tissue types (b) can be used to calculate shape, size and other statistics, and can also be subdivided into smaller regions to determine the amount of each tissue type in each lobe (c).

Figure 2

Cortical thickness maps. (a) An in vivo average cortical-thickness map created from 45 normally developing children at their first scan [12]. The brain surface is color coded according to the bar on the right, where thickness is shown in millimeters. The average thickness map can be compared to an adapted version of the 1929 cortical thickness map of von Economo [15] (b). Color coding has been applied over his original stippling pattern, respecting the boundaries of his original work, to highlight the similarities between the two maps. Reproduced, with permission, from [12].

Figure 3

Annualized rate of change in cortical thickness. The average rate of change in cortical thickness is shown in millimeters according to the color bar on the right (maximum gray-matter loss is shown in shades of red and maximum gray-matter gain is shown in shades of blue). Forty-five children were studied twice (two-year scan interval) between 5 and 11 years of age. Reproduced, with permission, from [12].

Figure 4

Brain-behavior maps for vocabulary and cortical thickness. P values are for negative correlations between change in cortical thickness (time 2 minus time 1, as shown in figure 2 of [12]) and change in vocabulary raw scores (time 2 minus time 1). Negative P values (i.e. regions where greater thinning was associated with greater vocabulary improvement) are color-coded, and regions in white showed no significant association. Positive correlations were not significant in the permutation analyses for any of the regions of interest, and are not shown here. Reproduced, with permission, from [12].

Figure 5

Mapping brain change over time. Brain changes in development can be identified by fitting time-dependent statistical models to data collected from subjects cross-sectionally (i.e. across a group of subjects at a particular time), longitudinally (i.e. following individual subjects as they aged), or both. Measurements such as cortical thickness are then plotted onto the cortex using a color code. (a,b) Trajectory of gray-matter loss over the human lifespan, based on a cohort of 176 subjects aged 7–87 years [11]. Plots superimposed on the brain in (b) show how gray-matter density decreases for particular regions; (a) highlights example regions in which the gray-matter density decreases rapidly during adolescence (the superior frontal sulcus) or follows a more steadily declining time-course during lifespan (the superior temporal sulcus). (c,d) Trajectory of cortical gray-matter density in 13 children scanned longitudinally every two years for eight years [4]. The units used in (a) and (d) are gray-matter density, which is defined as the proportion of tissue segmenting as gray matter within a 15-mm-diameter sphere centered at each point on the cortical surface. This widely-used measure ranges from 0 to 1, and is highly correlated with cortical thickness [52].

Figure 6

Tensor maps of growth and tensor-based morphometry. (a) The corpus callosum (indicated by a green box) of a healthy three-year-old girl in a sagittal section from a 3D MRI scan. Using a follow-up scan thee years later, an elastic deformation field is computed that digitally aligns, or warps, the anatomy of the earlier time-point to match its shape at the later time point. The amount of local stretching of the anatomy is color-coded, indicating fastest growth rates (red) in the anterior corpus callosum. In a related approach (b), maps can be compiled to represent the average expansion factor required to deform an average corpus callosum shape elastically onto each subject in a set of healthy children and matched autistic children. This identifies areas of the corpus callosum that are thinner in autistic children, pinpointing where abnormal white-matter growth might lead to the disorder in early childhood. The letters ‘S’ and ‘R’ denote the splenium and rostrum of the corpus callosum, respectively. Data in (a) is adapted, with permission, from [22]; (b) is adapted, with permission, from [53].

Figure 7

Differences in gray-matter density between subjects with three neurodevelopmental disorders. The percentage differences in gray-matter density between subjects with Williams syndrome (WS) (a), attention-deficit–hyperactivity disorder (ADHD) (b), fetal alcohol syndrome (FAS) (c) and their respective normally developing control groups are color-coded. In all maps, warmer colors represent positive differences, indicating an increase in the patient group (arbitrarily coded as 1) relative to the control group (arbitrarily coded as 0), with red representing the largest group difference. Note that the maximum value varies on the three color bars, depending on the maximum group difference from each comparison. Adapted, with permission, from [36,41,46].

Figure 8

Statistical maps of cortical structure. Various maps can be made that describe different aspects of cortical anatomy. These include maps of gyrus-pattern asymmetry (a) and how strongly genes and environmental influence brain structure (b). Panel (a) shows the increasing gyrus-pattern asymmetry in groups of children, adolescents and adults [55]. In these maps, the differences in asymmetry in the three age groups can be evaluated as the angle of the left relative to the right Sylvian fissure in each age group. Note the angle is larger in the adults than in the children. The color coding enables the displacement in millimeters between the left and right hemisphere within each age group to be visualized. Asymmetry measures can also be extended to rest of the cortical surface and expressed in millimeters. Panel (b) shows correlations in gray matter for groups of identical (monozygotic, MZ) and fraternal (dizygotic, DZ) twins. Some brain regions develop under tight genetic control, and these include many frontal and temporal lobe regions, such as the dorsolateral prefrontal cortex and temporal poles (red). Other regions are more strongly influenced by the environment as they develop (i.e. a greater proportion of the inter-subject variance is explained by non-genetic factors). (c) Average maps of gray-matter loss rates for healthy boys and girls, scanned longitudinally over five years. Also shown are maps of the considerably faster loss rates in age-matched and gender-matched subjects with childhood-onset schizophrenia scanned at the same ages and intervals. The frontal cortex underwent a selective rapid loss of gray matter (up to 3–4% per year faster in patients than controls). These changes might be, in some respects, an exaggeration of changes that normally occur in adolescence [10,54]. By contrast, deficits occurring as Alzheimer’s disease progresses are shown (d) by comparing average profiles of gray-matter density between 12 Alzheimer’s disease patients (mean age 68.4±1.9 years) and 14 age-matched controls (mean age 71.4±0.9 years). In Alzheimer’s disease, gray-matter loss sweeps forward in the brain from limbic to frontal cortices in concert with cognitive decline, but in the schizophrenia patients (c), the frontal cortices lose gray matter the fastest. The letters ‘W’ and ‘T’ denote Wernicke’s area and the temporal cortex, respectively.