Edited by: Emi Takahashi, Boston Children's Hospital, United States
Reviewed by: Harushi Mori, University of Tokyo, Japan; Kristina Aldridge, University of Missouri, United States; Ricardo Insausti, Universidad de Castilla-La Mancha, Spain
*Correspondence: Gentaro Taga
This article was submitted to Evolutionary Psychology and Neuroscience, a section of the journal Frontiers in Neuroscience
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The topographic relationships between the macroanatomical structure of the lateral cortex, including sulci and fissures, and anatomical landmarks on the external surface of the head are known to be consistent. This allows the coregistration of EEG electrodes or functional near-infrared spectroscopy over the scalp with underlying cortical regions. However, limited information is available as to whether the topographic relationships are maintained in rapidly developing infants, whose brains and heads exhibit drastic growth. We used MRIs of infants ranging in age from 3 to 22 months old, and identified 20 macroanatomical landmarks, featuring the junctions of major sulci and fissures, as well as cranial landmarks and virtually determined positions of the international 10-20 and 10-10 systems. A Procrustes analysis revealed developmental trends in changes of shape in both the cortex and head. An analysis of Euclidian distances between selected pairs of cortical landmarks at standard stereotactic coordinates showed anterior shifts of the relative positions of the premotor and parietal cortices with age. Finally, cortical landmark positions and their spatial variability were compared with 10-10 landmark positions. The results indicate that variability in the distribution of each macroanatomical landmark was much smaller than the pitch of the 10-10 landmarks. This study demonstrates that the scalp-based 10-10 system serves as a good frame of reference in infants not only for assessing the development of the macroanatomy of the lateral cortical structure, but also for functional studies of cortical development using transcranial modalities such as EEG and fNIRS.
Recent advances in neuroimaging methods have begun to shed light on developmental changes in cortical structures, which could, in turn, serve as a referential framework for describing functional development. Studies using magnetic resonance imaging (MRI) have revealed the development of the macroanatomy of the brain during fetal (Huang et al.,
There are ongoing efforts to establish large-scale infant MRI datasets (Almli et al.,
While the aforementioned studies have developed methods for predicting 10-20 standard electrode and/or fNIRS channel locations relative to the underlying macroanatomical structure of the cortex for different ages, few studies have focused on the fundamental issue of how the macroanatomy of the cortex and the skull co-develop and how individual and developmental variations are quantified throughout development. Aldridge et al. (
In this study, we asked how the brain and head co-develop during the first 2 years of life and whether the head coordinate system provides a good frame of reference not only for the study of the structural development of the brain, but also for the functional study of brain development using EEG and fNIRS in this age period. Thus, we sought a multi-facetted solution for this problem. We focused on several macroanatomically distinct cortical structures on the lateral cortical surfaces of infants between the ages of 3 and 22 months old. These structures were mainly selected at junctions of major sulci and fissures, which can be detected even in low-contrast MR images. Next, we examined topographic changes of the locations of these macroanatomically distinct structures during infant development. We attempted two different approaches for this examination. First, we performed a Procrustes analysis (Bookstein,
The MRI data of normally developing infants were obtained from the MRI data set previously reported (Tanaka et al.,
For the present study, acquired MRI data were aligned to the anterior commissure (AC) and posterior commissure (PC), with the AC-PC midpoint providing the origin, the AC-PC line forming the y-axis, and a midline forming the z-axis. The x-axis ran left to right. Skull stripping was performed using the automated brain extraction tool (BET; Smith,
Using the MRIcron software package (Rorden et al.,
Vertex: We determined the highest axial slice in which the parietal cortex appeared, and recorded the coordinates of the center-of-gravity of the parietal cortex in that axial slice.
Frontal pole: We determined the coronal slice in which the frontal cortex appeared, and recorded the coordinates of the center-of-gravity of the frontal cortex in that coronal slice.
Temporal pole: We determined the most anterior coronal slice in which the temporal cortex appeared, and recorded the coordinates of the center-of-gravity of the temporal cortex in that coronal slice.
Base of the brain: We determined the most ventral axial slice in which the temporal cortex appeared, and recorded the coordinates of the center-of-gravity of the temporal cortex in that axial slice.
Occipital pole: We determined the most posterior coronal slice in which the occipital cortex appeared, and recorded the coordinates of the center-of-gravity of the occipital cortex in that coronal slice.
Leftmost and rightmost points: We determined the leftmost and rightmost sagittal slices in which the temporal cortex appeared, and recorded the coordinates of the center-of-gravity of the temporal cortex in the sagittal slice for each side.
Upper edge of the central sulcus: We identified the central sulcus, and found the most dorsal axial slice where the posterior bank of the central sulcus could be determined. We recorded the most medial coordinates of the posterior bank on that axial slice.
Junction of the superior frontal sulcus and the precentral sulcus: We identified the superior frontal sulcus and the precentral sulcus, and found the junction of these sulci that appears on the axial slices. Hereafter, we refer to an intersection of two sulci as a “junction” according to Destrieux et al. (
Junction of the inferior frontal sulcus and the precentral sulcus: We identified the inferior frontal sulcus, and found the coronal and sagittal slices where the junction of this sulcus and the precentral sulcus appeared. We also determined the neighboring point of the junction on the lateral surface using the axial and coronal slices.
Root of the ascending ramus of the Sylvian fissure: We identified the Sylvian fissure and the ascending ramus of the Sylvian fissure. We found the root of the ramus on the most lateral (sagittal) slice. The neighboring point of this landmark on the lateral surface was also determined using the axial and coronal slices.
The inferior termination of the central sulcus (CS tip): We determined the inferior termination of the central sulcus that appears on the axial slice and recorded the coordinates of the most lateral point. The central sulcus sometimes connects with the Sylvian fissure. In this case, the junction of the central sulcus and the Sylvian fissure was considered the inferior termination of the central sulcus.
Junction of the posterior central sulcus and the intraparietal sulcus: We identified the posterior central sulcus and the intraparietal sulcus, and found the junction of these sulci on the axial and sagittal slices. We also determined the neighboring point of the junction on the lateral surface using the coronal slice.
Preoccipital notch: We identified the axial and sagittal slices where the preoccipital notch was most visible. We sometimes used a surface reconstruction of the MRI data to confirm the position. The most ventral and lateral coordinates were recorded.
Calcarine fissure: We identified the calcarine fissure using the axial and sagittal slices, and determined the most medial point where the fissure appeared on the lateral surface around the occipital pole.
Parietooccipital sulcus: We identified the parietooccipital sulcus on the sagittal slice and recorded the coordinates of the most dorsomedial point of the sulcus, which was observed on axial and sagittal slices.
Sylvian fissure on the coronal slice of CS tip: We determined the most lateral point of the Sylvian fissure on the coronal slice of the y coordinate of landmark
Superior temporal sulcus on the coronal slice of CS tip: We identified the superior temporal sulcus and determined the most lateral point of the sulcus on the coronal slice of the y coordinate of landmark 11.
Inferior temporal sulcus on the coronal slice of CS tip: We identified the inferior temporal sulcus, and determined the most lateral point of the sulcus on the coronal slice of the y coordinate of landmark 11.
The most ventral point of the lateral temporal cortex on the coronal slice of CS tip: On the coronal slice of the y coordinate of landmark 11, we determined the most ventral point of the lateral temporal cortex (This point was only used for analyzing relationship between 10-10 positions and macroanatomical landmarks).
Root of the posterior ascending branch of the Sylvian fissure: We determined the root of the posterior ascending branch of the Sylvian fissure on the sagittal slice, where the inclination of the Sylvian fissure changed drastically, and recorded its coordinates and the coordinates of the neighboring point on the lateral surface using axial and coronal slices.
The landmarks on a brain surface of a representative infant. We defined 20 landmarks on each hemisphere of the cerebral cortex. The red dots (from No. 1–6) are extreme points, including poles, in antero-posterior, ventro-dorsal, and left-right directions. The red dots (from No. 7–20) are cortical landmarks (see Section Materials and Methods). We delineated all major sulci of a 12-month-old infant in our previous study (Matsui et al.,
In a previous study, 12 landmarks distributed over the cortical surface were used to assess shape difference of the normal adult brain (Free et al.,
To analyze holistic variations in the shapes of the cortex and scalp using the Procrustes method (Bookstein,
Basically, Euclidian distances between given macroanatomical landmarks were used to assess their topological relationship (indicated as “direct distance” in Table
Topological relationship among macroanatomical cortical landmarks using Euclidian distances.
7. Upper end of the central sulcus | L | (2-7)/(7-5) | Projected to the sagittal plane | −0.034 | 0.900 | Unchanged along anterior-posterior axis |
R | −0.018 | 0.949 | ||||
8. Junction of the superior frontal sulcus and the precentral sulcus | L | (2-8)/(8-5) | Direct distance | −0.620 | 0.010 | Relative position around the upper precentral sulcus shifts anteriorly |
R | −0.570 | 0.021 | ||||
9. Junction of the inferior frontal sulcus and the precentral sulcus | L | (2-9)/(9-5) | Direct distance | −0.725 | 0.002 | Relative position around the lower precentral sulcus shifts anteriorly |
R | −0.742 | 0.001 | ||||
10. Root of the ascending ramus of the Sylvian fissure | L | (2-10)/(10-5) | Direct distance | −0.594 | 0.015 | Relative position of the ascending rami of the Sylvian fissures shifts anteriorly |
R | −0.555 | 0.026 | ||||
16. Lower (imaginary) root of the central sulcus | L | (2-16)/(16-5) | Direct distance | −0.579 | 0.019 | Significant anterior shift |
R | −0.562 | 0.023 | ||||
12. Junction of the posterior central sulcus and the intraparietal sulcus | L | (2-12)/(12-5) | Direct distance | −0.677 | 0.004 | Relative position of the junction of the posterior central and the intraparietal sulci shifts anteriorly |
R | −0.705 | 0.002 | ||||
15. Parieto-occipital sulcus | L | (2-15)/(15-5) | Projected to the sagittal plane | −0.447 | 0.082 | Moderate tendency of anterior shift of relative position of the parieto-occipital sulcus |
R | −0.420 | 0.105 | ||||
20. Root of the ascending brunch of the posterior Sylvian fissure | L | (2-20)/(20-5) | Direct distance | −0.397 | 0.127 | Moderate tendency of anterior shift of relative position around the supramarginal gyrus |
R | −0.417 | 0.108 | ||||
13. Preoccipital notch | L | (2-13)/(13-5) | Direct distance | −0.177 | 0.512 | Relative position of the preoccipital notch remain unchanged along anterior-posterior axis |
R | −0.253 | 0.345 | ||||
14. Calcarine fissure height | L | (1-14)/(14-4) | Projected to the coronal plane | −0.768 | 0.000 | Significant dorsal shift |
R | −0.659 | 0.006 | ||||
20. Sylvian fissure height | L | (1-20)/(20-4) | Projected to the coronal plane | 0.189 | 0.482 | Unchanged |
R | 0.284 | 0.286 | ||||
17. Superior temporal sulcus height | L | (1-17)/(17-4) | Projected to the coronal plane | −0.301 | 0.258 | Unchanged |
R | 0.356 | 0.177 | ||||
18. Inferior temporal sulcus height | L | (1-18)/(18-4) | Projected to the coronal plane | 0.327 | 0.216 | Unchanged in the left, while significant lowering in the right hemisphere |
R | 0.524 | 0.037 |
As a relative head-surface-based positioning system, we utilized the international 10-10 system (Chatrian et al.,
We analyzed the spatial variability of the macroanatomical cortical landmarks against the 10-10 landmarks. In infant head space, the relative location of a given point on the cortex (CP) can be expressed as a composition of vectors that refers to neighboring standard points of the infant's head surface such as the 10-10 landmarks (Okamoto and Dan,
Then, we obtained the corresponding locations of the selected macroanatomical landmarks on the 12-month-old infant template (Matsui et al.,
Thus, the selected macroanatomical cortical landmarks of an infant brain were transferred to the 12-month-old infant atlas space. However, since these points are located around the cortical surface of the 12-month-old infant brain, they were back-projected onto the scalp by linearly enlarging the three coefficients. In this way, all the macroanatomical cortical landmarks of the 15 infants were transformed to the scalp of the 12-month-old infant template in reference to the 10-10 landmarks.
Detailed expression of linear algebraic representations of cortical anatomy with cranial landmarks.
The infant brains used in the current study were aligned to a coordinate system with the AC-PC midpoint as the origin (Figure
Growth of the infant brain. We have set the AC-PC midpoint of each brain as the origin of the 3-D coordinates, and overlaid the left lateral views of three brains of 3- (brown), 10- (blue), and 18-month-old (green) infants. Note the elongation of the brain in the anterior-posterior direction (y-axis).
To examine changes in the shape of the cortex and head with age, positions of cortical, and head landmarks were analyzed using Procrustes analysis. Figure
Procrustes analysis of the cortical and head landmarks. Positions of landmarks of individuals and averaged positions of landmarks with the direction of the first principal component of variations are separately shown for the cortex and the head superimposed on the Procrustes coordinates.
First principal component scores as a function of age.
The results of analyses of the topological relationship among macroanatomical landmarks on the lateral cortical surface of infant brains are summarized in Table
Macroanatomical landmarks of 14 infants were transformed to the scalp surface of a 12-month-old infant based on their relative relation with three neighboring 10-10 positions, and were depicted in reference to the 10-10 positions (Figures
Relationship between 10-10 positions and macroanatomical landmarks. 10-10 landmarks (black dots) are depicted on the 12-month-old infant scalp template. The macroanatomical cortical landmarks from 14 infants, transformed using three neighboring 10-10 landmarks and projected onto the scalp, are visualized in different colors: 1, sky gray; 2, orange; 3, navy blue; 4, cherry pink; 5, magenta; 6, grass green; 7, violet; 8, malachite green; 9, brown; 10, wine red; 11, yellow; 12, olive green; 13, silver gray; 14, terracotta; 15, khaki; 16, burnt sienna; 17, white; 18, cobalt green; 19, blue and 20, red, where the numbers indicate the macroanatomical structures depicted in Table
Distribution of macroanatomical cortical landmarks projected on the scalp. Left and right macroanatomical cortical landmarks of 14 infants were transformed and projected onto the 12-month-old infant scalp template. Distances from their centroids are depicted using box-and-whisker plots where the horizontal red bars indicate the median, boxes represent upper and lower quartiles, and the upper and lower whiskers correspond to the smallest distance above and the largest distance below 1.5 inter-quartile range from the upper and lower quartiles. In the rightmost panel, distribution of the nearest distance from each 10-10 landmark is also depicted in a box-and-whisker plot as described above.
To confirm the robustness of the macroanatomical cortical landmarks, centroids for each landmark were projected onto the 12-month-old scalp template with delineated gyri as described in Matsui et al. (
Inter-subject centroids of the macroanatomical cortical landmarks on delineated gyri of the 12-month-old template. The delineated cortical gyri are projected onto the scalp (Matsui et al.,
The present study describes developmental changes in the topographical relationships among macroanatomical landmarks on the lateral cortical surface of developing infants between the ages of 3 and 22 months. The Procrustes analysis, focusing on holistic trends in relative distribution among the macroanatomical cortical landmarks, revealed a principal component represented by a bilateral posterior shift of the upper region around the central sulci and a left-lateral anterior shift of the region around the posterior edge of the Sylvian fissure as infants develop. In addition, the Procrustes analysis performed with the presence of 10-20 landmarks on the scalp revealed principal component represented by an elongation of the scalp landmark positions along the anterior-posterior axis. On the other hand, analyses of the topological relationships among macroanatomical landmarks from stereotactic coordinates indicated several developmental changes in relative positions among the cortical landmarks including the root of the central sulci, the root of the ascending branches of the posterior Sylvian fissures, the root of the ascending rami of the Sylvian fissures, the posterior root of the inferior and superior frontal sulci and the calcarine fissures. Despite the presence of developmental changes in relative locations of the cortical landmarks when expressed in stereotactic coordinates, their variability was rather small compared to cortical regions defined by the international 10-10 system. In the following discussion, we will examine the importance of these findings from developmental and technical points of view, and also present perspectives on how these findings can be practically implemented in fNIRS and EEG studies on infants.
The sample size of 16 participants ranging in age from 3 to 22 months in the present study was not enough to perform a group analysis. However, significant regressions of individual shape variables of cortical and scalp landmarks on age suggest that the present analysis captured the development of shape changes. By using the same analytical method, ontogenetic shape changes were assessed for human cranium during the fetal period (Morimoto et al.,
The Procrustes analysis performed on the cortical macroanatomical landmarks extracted a principal component that exhibited significant age-dependent change. This change seems to represent a dynamical topological change around the parietal lobe. More specifically, the upper central sulci moved posteriorly and the posterior part of the Sylvian fissure moved anteriorly. As shown in Figure
On the other hand, the Procrustes analysis of 10-20 landmarks on the scalp revealed a distinct principal component with posterior points shifting outward and temporal points shifting inward. These correspond to the relative elongation of the scalp along the anterior-posterior axis and the narrowing of the scalp on both sides. Since Euclidian distance analysis revealed significant elongation of the brain on the y-axis, while changes in brain size on the x-axis for this age range were smaller than those on other axes (Table
Concerning the general shape of the brain, there was an age-dependent increase in brain size in all directions in both hemispheres. However, the rate of enlargement differed among directions with the brain growing the most rapidly in the anterior-posterior axis, while growth was less eminent in the width and height directions. It should be noted that the evidence of the leftward occipital and rightward frontal asymmetry, known as petalia or Yakovlevian torque (LeMay,
Concerning macroanatomical structures on the lateral cortical surfaces, there were several obvious changes. The ventral root of the central sulcus, as measured at its imaginary cross fissure, shifted anteriorly in both hemispheres while such change was not observed in the dorsal edge of the central sulci. This change suggests that the ventral area of the frontal lobe moved forward as the brain elongated anteriorly in the course of development. This trend was also supported by the relative anterior shift of two macroanatomical landmarks in the of the ascending rami of the Sylvian fissure and the posterior root of the inferior frontal sulci in both hemispheres. Further, this global change in the frontal area seemed to be associated with a widening of the precentral gyri because the posterior root of the superior frontal sulci moved forward while the dorsal edge of the central sulci remained unmoved as infants developed.
Another macroanatomical change was found in the anterior roots of the intra-parietal sulci shifting anteriorly. Given other observations that relative positions of the parieto-occipital sulci remain rather stable, the anterior shift of the intraparietal sulci may suggest a relative enlargement of the border regions of the superior and inferior parietal lobules in the anterior direction. The other obvious change was the upward shift of the calcarine fissure on both hemispheres. Thus, the occipital lobe also underwent developmental changes. However, changes in the temporal lobe were not as pronounced. The only exception was a relative downward shift of the right inferior temporal sulcus while the left side remained unaffected.
These tendencies in relative changes in macroanatomical landmark positions were mostly in line with the results of macroanatomical delineation performed by Li et al. (
Given these analyses, we postulate that macroanatomical landmarks on the lateral cortical surfaces are useful indicators for describing infant brain development. Even with recent advancements in tissue segmentation techniques for low-contrast infant brain images, skull stripping to extract the infant cortical surface is difficult. The infant MRI scans used for the current study did not have sufficient image contrast to undergo automatic tissue segmentation, but were clear enough to allow detection of macroanatomical landmarks defined by orientations of major sulci.
In comparison with approaches using a standard brain template produced based on MRIs, a scalp-based positioning system, such as the international 10-20 system and its derivative the 10-10 system, would serve as a simple and robust approach to providing a common referential framework for the lateral cortical structure, even when the quality of an MRI is not sufficient for skull-striping, segmentation, and normalization. Thus, we explored the relationship between the 10-10 landmarks and macroanatomical cortical landmarks. The cortical landmarks on each infant brain were back-projected to the scalp and expressed in reference to three neighboring 10-10 positions. They were transformed to the scalp template of a 12-month-old infant, who also served as a template for extensive analyses in our former study (Matsui et al.,
The reliability of the 10-20 system to predict underlying cortical gyral structures has been established for adults (Okamoto et al.,
The current study demonstrates that macroanatomically distinct cortical structures on the lateral cortical surfaces defined as junctions between major sulci and fissures could serve as useful landmarks for the examination of topological features of brains between birth and 2 years of age. Their detection does not require high-resolution MRIs, and, thus, they serve as robust measures to describe macroanatomical changes in infant brains. A Procrustes analysis detected an age-dependent global trend manifesting as a posterior shift of the upper region around the central sulci and an anterior shift of the region around the posterior edge of the Sylvian fissure. Analyses of relative Euclidean distances among the macroanatomical landmarks revealed general shape differences as well as several distinct regional topological changes, most obvious of which were the forward shift of the macroanatomical structures on the prefrontal cortex and the parietal lobules as infants developed. Importantly, developmental changes in the relative topological orientation of the macroanatomical cortical landmarks were found to be sufficiently smaller than the area defined by the international 10-10 system. Therefore, we propose that the international 10-10 system can serve as a robust referential framework for positional descriptions in fNIRS study on infants.
This study was carried out in accordance with the recommendations of the Research and Ethics Committee at the University of Toyama with written informed consent from all subjects.
ID, FH, GT, and DT wrote the manuscript. MM provided all the primary data and the research environment. DT performed the majority of computational analyses image processing. GT performed the Procrustes analyses. FH and ID performed anatomical analyses. HW arranged research facilities and managed the research schedule. All authors contributed to conceiving research ideas, and interpretation of the results.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer HM declared a shared affiliation, though no other collaboration, with several of the authors DT, GT, and HW to the handling Editor, who ensured that the process nevertheless met the standards of a fair and objective review.
We thank Keiko Hirano and Tomoko Yoneyama for providing administrative assistance. We appreciate Michio Takahashi, Mitsuhiro Nakashima, and Satoshi Uda for providing assistance in data analysis.