A Preliminary Analysis of Sexual Dichotomy in the External Globus Pallidus N. van Berendonk University of Amsterdam Student nr: 10910972 Supervisor: dr. A. Alkemade Number of words: 5001 Abstract: 119
Abstract
Recent advancements in delineation of subcortical structures using ultra-high field magnetic resonance imaging (UHF MRI, 7T and higher) allow for clearer
visualization of the border between the internal and external globus pallidus (GPi/e), which is difficult at field strengths below 7T. Using 7T MRI, this study investigated sexual dichotomy in GPe volumes and locations, delineated by two independent raters on quantitative susceptibility maps (QSM) and T1-maps, in 23 healthy adults. The main analysis showed larger distances between left and right GPe in men, but no sexual dichotomy in volumes. Because of low inter-rater reliability, explorative
analyses were performed on the most reliable delineations, which also showed no sexual dichotomy in volumes, but larger distances between structures in men.
Introduction
To better understand human cognitive function, both cortical and subcortical networks must be explored, as motor, limbic, and cognitive networks span both levels (Keuken, Isaacs, Trampel, van der Zwaag, & Forstmann, 2018). Mapping the brain is a crucial step in this process; understanding the anatomy of components and their
connections in fundamental to understanding the system as a whole. Several human brain atlases are currently available for research purposes, such as the Harvard-Oxford atlas, the FreeSurfer atlas, and the Jülich cytoarchitectonic maps (Forstmann, de Hollander, van Maanen, Alkemade, 2017; Keuken et al., 2014). However, current atlases lack detailed information on the subcortex, as they contain only 7% of the estimated 455 subcortical structures that make up for approximately 25% of total human brain volume (Alkemade, Keuken, & Forstmann, 2013; Forstmann et al., 2017; Keuken et al., 2014, 2018). The main reason for this lack of detailed information is that most atlases are based on MR imaging at
field strengths up to 3 Tesla (3T). At these field strengths, delineation of subcortical
structures is challenging, since these structures are small and located deep in the brain at close proximity, and signal-to-noise ratio (SNR) decreases for structures deep inside the brain due to the increased distance from the MRI coil, resulting in lower anatomical specificity (De Hollander, Keuken, & Forstmann, 2015).
However, ultra-high field (UHF, 7T and higher) MRI provides a higher SNR and allows detailed visualization of individual subcortical structures (Forstmann et al., 2017; De Hollander et al., 2015). Keuken et al. (2014) introduced a new subcortical probabilistic atlas based on 7T MRI, in which they included the basal ganglia, consisting of the striatum
(putamen and caudate nucleus), the internal and external globus pallidus (GPi/e), the subthalamic nucleus (STN), and the substantia nigra pars reticulata (SNr). Today, around 169 subcortical structures have been visualized with 7T MRI (Keuken et al., 2018),
demonstrating that with 7T MRI it is possible to capture the anatomical variability of individual structures of the subcortex, including the basal ganglia.
Basal Ganglia
The basal ganglia (BG) are bilaterally located deep in the brain and are part of several cortico-subcortical circuits, including motor, prefrontal, associative and limbic circuits. Input to the basal ganglia comes from several cortical regions. The striatum and STN are the main input nuclei, and the GPi and SNr are the main output nuclei, projecting to the thalamus and brainstem. The thalamus projects back to the cortex. The motor circuit of the BG is most extensively studied, since a disturbance in BG output is known to
generate movement disorders including Parkinsonism (Middleton & Strick, 2000; DeLong & Wichmann, 2007). The BG are traditionally described to have two distinct pathways, direct and indirect, based on a segregation between inhibitory and excitatory neurotransmission in
the striatum. The direct pathway generates an increase in movement, through inhibitory influences on the thalamic neurons from the SNr and GPi. The indirect pathway leads to a reduction of movement, by excitatory influences on the thalamus through inhibition of the GPe and disinhibition of the STN. However, BG pathways may be more interrelated, both functionally and structurally (Calabresi, 2014).
Globus Pallidus
The GP (Figure 1) is located medial to the putamen and inferior to the caudate nucleus and consists of two distinct layers, separated by the lamini pallidi medialis (Lenglet
et al., 2012).
Figure 1
Location of the GP in sagittal (left), coronal (center) and axial (right) view.
The GPi is well documented as a target for deep brain stimulation (DBS), in order to relieve movement disorders in Parkinson’s disease (PD) (Odekerken et al., 2013). Jaeger and Kita (2011) suggest distinct functional roles for the two GP segments, based on their distinct connectivity. While the GPi is a BG output nucleus, the GPe has multiple
feed-synaptic input. Input to the GP comes from the striatum, STN, thalamic and brainstem nuclei. Additionally, the GPe provides input to the GPi. Striatal input is inhibitory, while STN input is excitatory to the GP. The striatum contains two groups of projecting neurons; one group that projects to the GPe, GPi and the SNr, and another group that projects
exclusively to the GPe. The GPe itself also shows a dichotomous cellular organization that supports its involvement in distinct pathways (Mallet et al., 2012). Considering the distinct and central connectivity of the GPe within BG circuits, further exploration of this structure is necessary, since its exact role is still unclear.
Sexual Dichotomy
To further explore the functional role of the GPe, a reliable anatomical delineation that allows researchers to distinguish between the two parts of the GP, while taking into account individual variability, is needed. Though general human brain anatomy is similar, there is large individual variability at both cortical and subcortical level, related to factors such as genetics, gene-environment interactions, (healthy) aging, sex, and disease (Keuken et al., 2013; 2015). Individual variability at the cortical level has been widely studied, but research at subcortical level only recently started advancing, due to improved imaging possibilities provided by UHF MRI. These studies primarily explore general
variability factors such as age and sex in a healthy population, as understanding normal variability is key to identifying anomalies (Keuken et al., 2018).
Several sexual dimorphisms in global brain anatomy have been well documented, including that men have larger overall brain volumes, larger lateral ventricles, and higher percentages of white matter (WM) and cerebrospinal fluid (CSF), while women have higher percentages of gray matter (GM) (Rijpkema et al., 2011). Region-specific sex differences in cortical thickness and WM-GM ratio are reported (Cosgrove, Mazure, & Staley, 2007;
Sowell et al., 2007), and sex differences in regional GM concentrations may change with age (Takahashi, Ishii, Kakigi, & Yokoyama, 2011), although this effect is not consistently found (Smith, Chebrolu, Wekstein, Schmitt, & Markesberry, 2007).
At the subcortical level, age as a factor for individual variation is much better documented than sex (Keuken et al., 2017). Indicated subcortical sex differences are that the amygdala and hypothalamus are larger in men, while women have a larger caudate nucleus and hippocampus (Cosgrove et al., 2007), although others report that the GP and putamen are larger in men (Rijpkema et al., 2011). Several studies indicate effects of age and age-sex interactions, such as structural differences in development during adolescence (Raznahan et al., 2014), and sex differences in age-related atrophy in the hippocampus, amygdala, BG and thalamus (Li et al., 2012). These cortical and subcortical sex differences indicate that sexual dichotomy is not general, but specific to certain regions or nuclei
(Rijpkema et al., 2011; Goldstein, 2001).
Investigating sexual dichotomy helps elaborate which variability factors are relevant for new neuroanatomical atlases at UHF MRI. In the present study, attention is focused on sexual dichotomy in the GPe, with two main objectives. The first is to contribute to mapping sex differences within the BG, as the GPe has not yet been studied individually in this regard. The second objective is to contribute to new neuroanatomical atlases, by providing delineations of the GPe to the collection of BG structures.
Sexual Dichotomy in the GP(e)
Previous research on sexual dichotomy in the GP is inconclusive, and no studies were found that looked at the GPe individually in this perspective. Some studies found no evidence of sexual dichotomy in the GP. Goldstein (2001) compared manually segmented GP volumes (cm3) between 48 healthy adults (21 females, mean age 39.8 years) and found
no difference between sexes. Volumes, obtained at 1.5T MRI, were relative to total cerebral volume. Gur, Gunning-Dixon, Turetsky, Bilker, and Gur (2002) compared absolute GP volumes (ml) between 116 healthy adults (59 females, ages 18 - 49 years) and reported slightly larger volumes for men, although not significant. Volumes were obtained by manual border-delineation at 1.5T MRI. Although no significant differences were found, these studies indicate that correcting for total cerebral volume may influence results. This is
further explored by Ahsan et al. (2007), who compared the GP volumes (mm3) of 30 healthy
adults (15 females, median age 31 years), and found that men have a 15% larger GP than women when compared in native space, and that this difference reduces to 5% when correcting for total hemispheric volume. Volumes were obtained by manual delineation at 1.5T MRI with isotropic voxel sizes of 0.9375 mm3.
Additionally, different field strengths also provided different results. Rijpkema et al. (2011) compared automatically segmented GP volumes (mm3) of 1004 healthy adults (ages
18 - 36 years). One group (n = 463, 261 females) was scanned at 1.5T and the other (n = 541, 330 females) at 3T, with isotropic voxel sizes of 1mm3, while controlling for total
GM/WM. They reported that at 1.5T, the GP appeared 3.9% larger for men, while at 3T, the GP appeared only 2.7% larger. The same effect was seen in other subcortical volumes, indicating the importance of higher resolution imaging techniques. Although these studies do not provide conclusive results about sexual dichotomy in the GP, they do indicate
relevant factors for further investigation, such as the correction for total cerebral volume and the importance of using higher field strengths.
Recent studies using UHF MRI have started investigating the GPe separately, but no sex differences have yet been reported as most studies focused on parcellation of the structure and potential age effects (Keuken et al., 2018). However, these studies provide good evidence for the use of specific MRI contrasts for delineation of the GPe, such as
T1-maps and quantitative susceptibility mapping (QSM). As T1 values correlate with myelin content in tissue, T1-maps allow visualization of contrasts between structures that vary in myelin density (Keuken et al., 2014). QSM is a novel type of post-acquisition susceptibility weighted imaging (SWI) that maps magnetic susceptibility values of tissue, and provides good visibility of structures rich in iron (Abosch, Yacoub, Ugurbil, & Harel, 2014; Deistung et al., 2013; Keuken et al., 2014). Since the GP has high levels of iron, QSM is ideal for
distinguishing between the GPi and the GPe, as it allows excellent visibility of the lamina pallidi medialis (Keuken et al 2014; Plantinga et al., 2014).
Although previous work regarding sex differences in the GP provides a good foundation for further research, some limitations are present. First of all, all
abovementioned studies looking into sex differences were done at field strengths of 1.5T to 3T. Possibly, due to the lack of anatomical specificity at these field strengths, certain
anatomical differences were imperceptible. Therefore, using 7T MRI to study sexual dichotomy in the GPe may wage different results. Second, none of these studies looked at the GPe individually. Considering the field strengths, this is understandable. However, it can not be assumed that when sex differences are found in the GP as a whole they also apply to the GPe individually, as evidence suggests different functional roles for the internal and external parts. It is therefore important that the GPe is studied individually. Finally, in previous research, methods for delineation vary between manual, automatic and semi-automatic, and various voxel sizes were used, with some studies not specifying their delineation methods or voxel sizes at all or incompletely, rendering them unsuitable for comparison and replication. This may also contribute to the inconsistency of their results.
In the current study, using 7T MRI, manual segmentation is done on contrasts that provide clearer visualization of the border between the GPi and GPe (QSM, T1-maps). Sexual dichotomy in the GPe is explored by comparing absolute GPe volumes and
locations between sexes. The main hypothesis tests for sexual dichotomy in GPe volumes; with the H0 that GPe volume is not sexually dichotomous; the H1 that men have larger GPe
volumes than women, and the H2 that women have larger GPe volumes than men. In
accordance with previously reported sex differences at global, cortical, subcortical and BG level, and in the GP, it is expected that the GPe will proof to be sexually dichotomous. Additionally, without any hypothesis, due to absence of previous research, sex differences in GPe location are explored by comparing distances between left and right GPe between sexes.
Methods Participants
Structural MR images from a total of 30 healthy adults were obtained from a larger pool of data (n = 100), collected by the Integrative Model-based Cognitive Neuroscience (IMCN) research unit, University of Amsterdam. Images of 15 male and 15 female participants from different age categories (19 - 69 years) were selected in randomized fashion. Participants who were not MRI-compatible were excluded. All included participants were informed about safety regulations, possible discomforts during scanning and
procedures regarding incidental findings, and signed informed consent forms previous to participation. Ethical approval was granted by the Ethics Review Board (ERB) from the faculty of Social and Behavioral Sciences, University of Amsterdam.
Data Acquisition
Images were made at the Spinoza Centre for Neuroimaging, Amsterdam, the Netherlands, using a Philips Achieva 7T MRI scanner with a 32-channel head array coil. Images were reoriented and converted from native Philips format (PAR/REC) to Nifti format
(Neuroimaging Informatics Technology Initiative) using Matlab. Images were made
anonymous by using a brain extraction tool (BET), removing all features except the brain. T1 and T2 contrasts were obtained using a MEMP2RAGE (multi-echo magnetization-prepared rapid gradient echo) sequence, an extension of the MP2RAGE sequence by Marques et al. (2010) (flip angles FA1,2 = [4°,4°]; TRGRE1,2 = [6.2ms, 31ms]; bandwidth =
404.9 MHz; TRMP2RAGE = 6778ms; acceleration factor SENSEPA = 2; FOV = 205 x 205 x
164mm; acquired voxel size = 0.70 x 0.70 x 0.70mm; acquisition matrix = 292 x 290mm; reconstructed voxel size = 0.64 x 0.64 x 0.70mms; turbo factor = 150, acquisition time 19.53 min; 176 shots). MP2RAGE is a T1-weighted structural scan, based on volumes with different inversion times acquired from two rapid gradient echo (GRE) images (Keuken et al., 2014). GRE images were acquired in sagittal plane after a 180°-degree inversion pulse and excitation pulses with inversion times TI1,2 = [670ms, 3675.4ms]. Multi-echo readouts
were added to the second inversion at four echo times (TE1 = 3ms, TE2,1-4 = 3, 11.5, 19,
28.5ms). T1-maps were computed using a look-up table (Marques et al., 2010). QSM images were computed using LSQR from pre-processed phase maps (using integrated phase unwrapping and background phase removal using the Laplacian; iHARPERELLA) (Li, Avram, Wu, Xiao, & Liu, 2014; Li et al., 2015).
Segmentation
Manual segmentations of the GPe were performed by 2 independent raters who previously completed segmentation training, and were supervised by neuro-anatomist A.A. Segmentations were done using QSM and T1-maps in FSL 4.1.4 viewer (Jenkinson,
Beckmann, Behrens, Woolrich, & Smith, 2012). Sagittal, coronal, and axial views were used interchangeably during segmentation, allowing the raters to cross-check their delineations. Voxel intensities were adjusted to optimize contrasts, using the BriCon option (-0.2 - 0.2 for
QSM and 0 - 4050 for T1-maps). For each brain, separate masks were made for the left and right GPe. To avoid bias, raters were advised to alternate the hemisphere in which they started and to finish both hemispheres in one session. Lighting and screen brightness settings were adapted depending on the preferences of each rater. To begin each
segmentation, anatomical landmarks were located to help distinguish the GPe within the brain, such as the caudate nucleus and the putamen. A starting point was chosen on a clear border of the structure, after which more border points were identified until the remainder of the mask could be filled in. Masks were made partially transparent to check border delineations during segmentations.
Analysis
Mask volumes were binarized by coding all colored pixels coded as “1” and all non-colored pixels as “0”. Next, the conjuncts of the masks were calculated by multiplying the binarized masks of the two raters, indicating overlap between two masks (in voxels and mm3). The conjuncts were used to calculate a dice similarity coefficient (DSC) (Dice, 1945). The DSC is a statistical validation metric, used to evaluate overlap accuracy and
reproducibility between manual segmentations of MR images (Zou et al., 2004). DSC was calculated for all conjunct-masks of left and right GPe separately, using the following formula:
𝐷𝐷𝐷𝐷𝐷𝐷 =2 ∗ |𝑚𝑚1 ∩ 𝑚𝑚2| |𝑚𝑚1| + |𝑚𝑚2|
Here, |𝑚𝑚1 ∩ 𝑚𝑚2| is the volume of the conjunct mask, with |𝑚𝑚1| and |𝑚𝑚2| representing the masks by Rater 1 and 2. A DSC of 1 indicates complete overlap, while a value of 0 indicates no overlap at all. For small nuclei, a DSC of around 0.75 is estimated to be
comparison to the total volume of the mask. For larger nuclei, a DSC of around 0.9 would be preferred. An example of a high DSC on segmentations of the GPe can be found in Keuken et al. (2014), who reached a mean DSC of 0.88 (SD = 0.01).
To compare GPe locations, the center of gravity (CoG) coordinates for each mask were obtained from FSL. For each participant, distances (mm) between left and right GPe were calculated using the Euclidian distance formula, with 𝑥𝑥2, 𝑦𝑦2 and 𝑧𝑧2 representing the CoG coordinates for the right GPe and 𝑥𝑥1, 𝑦𝑦1 and 𝑧𝑧1 representing the CoG coordinates for the left GPe:
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐷𝐷𝐸𝐸𝐷𝐷𝐷𝐷𝐸𝐸𝐸𝐸𝐸𝐸𝐷𝐷 = �(𝑥𝑥2 − 𝑥𝑥1)2+ (𝑦𝑦2 − 𝑦𝑦1)2+ (𝑧𝑧2 − 𝑧𝑧1)2
Statistical analyses were done using JASP 0.8.6.0 (http://jasp-stats.org). First, descriptive analyses of the mask volumes, conjunct-masks, DSC and Euclidian distance were performed. Next, dependent variables GPe-L conjunct and GPe-R conjunct were tested for sex differences using repeated measures ANOVA, while controlling for the influence of age and mean DSC as covariates. Finally, dependent variable distance between left and right GPe was tested for sex differences, using repeated measures ANOVA, while controlling for age and mean DSC as covariates. The threshold for significance was set at .05 for both analyses.
Results
Descriptive analysis provided 7 outliers (3 females, 4 males). All outliers were
individually inspected before removal (Appendix A); one was caused by a corrupt mask; the others were considered statistical outliers. New descriptive statistics were obtained for the remaining dataset. Mean ages and age range changed slightly for men (Table 1). Removal
of outliers resulted in lower mean GPe volumes in both hemispheres for both raters and lower conjunct volumes (Tables 2, 3 and 4).
Table 1
Population characteristics: number of participants (N), mean ages (standard deviations in parentheses), and age ranges in each group.
Age
With outliers Without outliers Age
f m f m N 15 15 12 11 Mean 40.33 (16.15) 40.47 (15.63) 40.50 (14.64) 38.82 (12.56) Range 19-69 19-69 19-66 19-58 Table 2
Mean volumes (mm3), standard deviations (in parentheses), and minimum and maximum volumes of the left GPe, for both raters.
GPe L Rater 1
With outliers Without outliers GPe L Rater 1 GPe L Rater 2 With outliers Without outliers GPe L Rater 2
f m f m f m f m Mean 1144.81 (186.45) 1220.67 (241.56) 1078.20 (125.14) 1179.17 (200.73) 517.98 (139.91) 498.21 (161.39) 500.48 (125.47) 419.26 (67.90) Minimum 813.38 877.52 813.38 877.52 358.08 318.59 358.08 318.59 Maximum 1589.08 1774.59 1266.38 1600.59 811.65 849.33 720.19 522.31 Table 3
Mean volumes (mm3), standard deviations (in parentheses), and minimum and maximum volumes of the right GPe, for
both raters.
GPe R Rater 1
With outliers Without outliers GPe R Rater 1 GPe R Rater 2 With outliers Without outliers GPe R Rater 2
f m f m f m f m Mean 1137.91 (220.92) 1204.86 (209.34) 1051.31 (143.08) 1174.80 (172.16) 498.08 (151.35) 528.05 (180.56) 481.97 (126.19) 441.79 (62.04) Minimum 865.44 831.21 865.44 941.66 273.81 358.52 273.81 358.52 Maximum 1524.37 1608.64 1294.85 1608.64 855.95 999.18 657.49 548.77 Table 4
Mean conjunct volumes (mm3), standard deviations (in parentheses), and minimum and maximum conjunct volumes of left
and right GPe, indicating overlap in masks.
Conjunct Volume GPe L With outliers Conjunct Volume GPe L Without outliers Conjunct Volume GPe R With outliers Conjunct Volume GPe R Without outliers f m f m f m f m Mean 504.90 (128.26) 481.72 (138.30) 488.47 (112.52) 413.79 (67.46) 477.33 (123.08) 513.60 (164.28) 466.49 (111.46) 436.96 (60.41) Minimum 358.08 318.59 358.08 318.59 273.81 357.95 273.81 357.95 Maximum 786.92 787.78 686.54 519.72 732.85 958.34 629.02 540.43
Mean DSC for left and right GPe became slightly higher for women and slightly lower for men, and ranges became smaller (Table 5). Overall DSC became slightly lower. Values
were below 0.6, both before and after removal of outliers, indicating poor inter-rater reliability (Table 6). Mean distance showed a slight increase in both sexes, for both raters (Table 7).
Table 5
Mean DSC, standard deviations (in parentheses), and minimum and maximum DSC of left and right GPe, indicating inter-rater reliability.
DSC GPe L
With outliers Without outliers DSC GPe L With outliers DSC GPe R Without outliers DSC GPe R
f m f m f m f m Mean 0.61 (0.12) 0.56 (0.11) 0.62 (0.11) 0.52 (0.06) 0.59 (0.14) 0.59 (0.11) 0.61 (0.13) 0.54 (0.05) Minimum 0.41 0.41 0.46 0.41 0.40 0.48 0.40 0.48 Maximum 0.80 0.79 0.80 0.60 0.78 0.84 0.78 0.62 Table 6
Overall DSC, standard deviations (in parentheses), and minimum and maximum DSC over both hemispheres for all participants, indicating total inter-rater reliability.
Overall DSC
With outliers Without outliers Overall DSC
Mean 0.58 (0.11) 0.57 (0.10)
Minimum 0.43 0.44
Maximum 0.82 0.79
Table 7
Mean distances (mm), standard deviations (in parentheses), and minimum and maximum distances between left and right GPe, for both raters.
Euclidian Distance Rater 1 With outliers Euclidian Distance Rater 1 Without outliers Euclidian Distance Rater 2 With outliers Euclidian Distance Rater 2 Without outliers f m f m f m f m Mean 36.05 (2.62) 38.06 (2.26) 36.23 (2.09) 38.92 (1.83) 35.01 (2.99) 37.00 (1.90) 35.28 (2.19) 37.74 (1.66) Minimum 31.24 33.55 33.23 36.24 28.74 33.59 31.54 35.57 Maximum 40.61 42.22 39.78 42.22 40.29 39.97 39.95 39.97
Repeated measures ANOVA on the effect of sex on GPe volume included model terms sex (between-subjects effect), age (covariate), mean DSC (covariate) and sex * age (interaction). Assumptions for sphericity and equality of variances were met (Mauchly’s W = 1.0; Levene’s F(1,21) = 1.113, p = 0.303 (left conjunct) and F (1,21) = 4.103, p = 0.056 (right conjunct)). A significant between-subjects effect was found for the effect of mean DSC on GPe volumes (MS = 150673.360, F(1,18) = 35.711, p <.001, 𝜂𝜂2 = 0.608). No
Figure 2
Mean conjunct-volumes of left and right GPe, with 95% confidence intervals.
Repeated measures ANOVA on the effect of sex on distances between left and right GPe included model terms sex (between-subjects effect), age (covariate) and mean DSC (covariate). Assumptions for sphericity and equality of variances were met (Mauchly’s W = 1.0; Levene’s F(1,21) = 0.263, p = 0.613 (Rater 1) and F(1,21) = 1.050, p = 0.474 (Rater 2)). A significant within-subjects effect of mean DSC on distance was found (MS = 2.064, F(1,19) = 9.215, p = 0.007, 𝜂𝜂2 = 0.269), as well as a significant between-subjects effect of
sex on distance (MS = 65.952, F(1,19) = 8.723, p = 0.008, 𝜂𝜂2 = 0.299), indicating that
distances between left and right GPe are larger in men, as shown in Figure 3.
Figure 3
Explorative analyses
Low DSC values between raters were explored by visual inspection of 12 GPe
masks with the lowest and highest DSC (no outliers). Mask volumes varied greatly between raters, with Rater 1 providing larger volumes than Rater 2, and the conjunct volumes for each GPe closely resembled the volumes of Rater 1 (Appendix A).
A paired samples T-Test (n = 23) was used to test whether Rater 1’s volumes were significantly larger than Rater 2’s (Table 9). Criteria for normality (Shapiro-Wilk) were met (GPe-L: W = 0.982, p = 0.939; GPe-R: W = 0.972, p = 0.737). Results were significant for both left and right GPe volumes (t = 15.385, df = 22, p <.001, Cohen’s d = 3.208, and t = 14.671, df = 22, p <.001, Cohen’s d = 3.059), confirming that masks of Rater 1 were significantly larger than those of Rater 2 (Figure 4a/b).
Figure 4
Difference between mask volumes of Rater 1 and 2 for the left GPe (a, left) and the right GPe (b, right), with 95% confidence intervals.
Table 9
Mean volumes (mm3) and standard deviations (in parentheses) of left and right GPe by Raters 1 and 2.
Volumes Rater 1 Volumes Rater 2
GPe-L GPe-R GPe-L GPe-R
Mean 1126.49 (169.72) 1110.37 (166.39) 461.63 (108.11) 462.76 (100.66)
To allow for interpretation of these differences between raters, GPe volumes from this study were compared to volumes reported by Keuken et al. (2014), who also performed manual segmentations on the QSM contrast and reached a mean DSC of 0.88 (SD = 0.02). This comparison (Appendix A) indicates that in the current study, Rater 1 provides the best approximation of true GPe volumes, which was confirmed by A.A.
The main analyses on sexual dichotomy in GPe volumes and locations were repeated separately for Rater 1. Repeated measures ANOVA was used to test for sex differences in GPe volumes, with Rater 1’s left and right mask volumes as dependent variables. Model terms included were sex (between-subjects effect), age (covariate) and sex * age (interaction). Assumptions for sphericity and equality of variances were met (Mauchly’s W = 1.0; Levene’s F(1,21) = 1.389, p = 0.252 (GPe-L) and F(1,21) = 0.072, p = 0.791 (GPe-R)). Although GPe volumes were slightly higher for men (Table 10), no
significant within- or between-subjects effects were found (Figure 5).
Figure 5
Differences in left and right GPe volumes between sexes for Rater 1, with 95% confidence intervals. Table 10
Rater 1’s mean volumes (mm3) and standard deviations (in parentheses) for left and right GPe in men and women.
Mean Volume
f m
GPe L 1078.20 (124.14) 1179.17 (200.73)
An independent samples T-Test was used to test for sex differences in GPe location, with distances between left and right GPe as the dependent variable. Criteria for normality (Shapiro-Wilk) were met (men: W = 0.957, p = 0.744; women: W = 0.974, p = 0.921). Mean distance was 38.92 mm (SD = 1.83) in men and 36.23 mm (SD = 2.09) in women (Figure 6). Results showed that distances between structures were significantly larger in men compared to women (t = -3.265, df = 21, p = 0.004, Cohen’s d = -1.363).
Figure 6
Mean distances between left and right GPe for Rater 1, with 95% confidence intervals.
Discussion
The main analysis provides no evidence of sexual dichotomy in volumes of the GPe. This is confirmative of the H0, but contradictory to the expectation GPe volumes would be
larger in men. The main analysis also shows larger distances between left and right GPe in men, indicating a sex difference in GPe location. Finally, an effect of mean DSC on GPe volumes and distances was found. This is explained by the fact that the main volume analysis compared conjunct volumes, which are part of the DSC calculation.
The main limitation of this study is the low inter-rater reliability, due to which results from the main analysis can not be interpreted in relation to the research question. Visual
inspection of segmentations showed large volume differences between the two raters (Appendix A). Explorative analysis showed that Rater 1’s volumes were significantly larger than Rater 2’s. Several areas of difficulty were identified, which were mainly related to delineation of borders with the GPi and the anterior commissure. This may be due to limited experience of the raters, although other raters in the same department, with similar
experience, have reached high DSC values. Comparison with volumes from Keuken et al. (2014) and visual inspection by A.A. indicated that in this study, Rater 1 is most likely to be reliable. In the effort of providing a preliminary conclusion on sex differences in GPe
volumes and locations, explorative analyses were performed on Rater 1’s segmentations, which provided no evidence for sexual dichotomy in volumes, but did find larger distances between left and right GPe in men.
Another limitation is that in the current study, absolute volume and distance were measured without controlling for total cerebral volume. Total volumetric differences between sexes may provide an explanation for the sex differences in distance found in both the main and explorative analyses. As previously reported, men have greater cerebral volume and thus may show greater distance between bilateral structures. Therefore, it may be relevant to add a measure of distance between structures to the further delineation of BG structures.
The absence of evidence for sex differences in GPe volumes is relevant to the further delineation of BG structures, as it confirms that sexual dichotomy is not a global effect. Other subcortical structures may still proof to be sexually dichotomous. A possible explanation for region-specific differences between sexes may be found in sex differences in concentrations of sex steroid receptors between structures throughout development (Rijpkema et al., 2011; Goldstein, 2001). Also, sexual dichotomy in other variables, such as connectivity and cellular anatomy, should be considered. However, this may require other
methods than UHF MRI, such as post mortem microscopy analyses (Alkemade et al., 2013).
Further research regarding the GPe may benefit from using more experienced raters and stricter segmentation protocols to reach higher inter-rater reliability. Such protocols are currently under development. It is relevant that the effect of sex is studied more extensively, since the current study has not been able to provide reliable results.
The current study only makes a small contribution to the large amount of work that needs to be done on the delineation of subcortical structures. Beside the number of other structures yet to be studied, there are also other factors of individual variability that need to be investigated. The preliminary conclusion of this work is that there is no sexual dichotomy in GPe volumes, but GPe structures are located further apart in men, indicating that when measuring GPe location without correcting for total cerebral volume, a separation between sexes is relevant for new anatomical atlases at the level of the GPe.
References
Abosch, A., Yacoub, E., Ugurbil, K., & Harel, N. (2010). An assessment of current brain targets for deep brain stimulation surgery with susceptibility-weighted imaging at 7 tesla. Neurosurgery, 67(6), 1745–1756. doi:10.1227/NEU.0b013e3181f74105
Ahsan, R. L., Allom, R., Gousas, I. S., Habib, H., Turkheimer, F. E., Free, S., … Hammers, A. (2007). Volumes, spatial extents and a probabilistic atlas of the human basal ganglia and thalamus. NeuroImage, 38(2), 261–270. doi:10.1016 /j.neuroimage.2007.06.004 Alkemade, A., Keuken, M. C., & Forstmann, B. U. (2013). A perspective on terra incognita:
uncovering the neuroanatomy of the human subcortex. Frontiers in Neuroanatomy, 7. doi:10.3389/fnana.2013.00040
Calabresi, P. (2014). Direct and indirect pathways of basal ganglia: a critical reappraisal.
Nature Neuroscience, 17(8), 1022–1030. doi:10.1038/nn.3743
Cosgrove, K., Mazure, C., & Staley, J. (2007). Evolving knowledge of sex differences in brain structure, function, and chemistry. Biological Psychiatry, 62(8), 847–855. doi:10.1016/j.biopsych.2007.03.001.
Deistung, A., Schäfer, A., Schweser, F., Biedermann, U., Turner, R., & Reichenbach, J. R. (2013). Toward in vivo histology: A comparison of quantitative susceptibility mapping (QSM) with magnitude-, phase-, and R2*-imaging at ultra-high magnetic field strength.
NeuroImage, 65, 299–314. doi:10.1016/j.neuroimage.2012.09.055
DeLong, M. R., & Wichmann, T. (2007). Circuits and circuit disorders of the basal ganglia.
Archives of Neurology. doi:10.1001/archneur.64.1.20
Dice, L. R. (1945). Measures of the Amount of Ecologic Association Between Species.
Forstmann, B. U., de Hollander, G., van Maanen, L., Alkemade, A., & Keuken, M. C. (2017). Towards a mechanistic understanding of the human subcortex. Nature
Reviews Neuroscience, 18(1), 57–65. doi:10.1038/nrn.2016.163
Goldstein, J. M. (2001). Normal Sexual Dimorphism of the Adult Human Brain Assessed by In Vivo Magnetic Resonance Imaging. Cerebral Cortex, 11(6), 490–497. doi:10.1093 /cercor/11.6.490
Gur, R. C., Gunning-Dixon, F. M., Turetsky, B. I., Bilker, W. B., & Gur, R. E. (2002). Brain Region and Sex Differences in Age Association With Brain Volume: A Quantitative MRI Study of Healthy Young Adults. The American Journal of Geriatric Psychiatry,
10(1), 72–80. doi:10.1097/00019442-200201000-00009
De Hollander, G., Keuken, M. C., & Forstmann, B. U. (2015). The subcortical cocktail problem; Mixed signals from the subthalamic nucleus and substantia nigra. PLoS
ONE, 10(3). doi:10.1371/journal.pone.0120572
Jaeger, D., & Kita, H. (2011). Functional connectivity and integrative properties of globus pallidus neurons. Neuroscience, 198, 44-53. doi:10.1016/j.neuroscience.2011.07.050 JASP Team. (2018). JASP (Version 0.8.6.0) [Computer software]. Retrieved from
http://jasp-stats.org
Jenkinson, M., Beckmann, C. F., Behrens, T. E. J., Woolrich, M. W., & Smith, S. M. (2012). FSL. NeuroImage, 62(2), 782–790. doi:10.1016/j.neuroimage.2011.09.015
Keuken, M. C., Bazin, P.-L., Schafer, A., Neumann, J., Turner, R., & Forstmann, B. U. (2013). Ultra-High 7T MRI of Structural Age-Related Changes of the Subthalamic Nucleus. Journal of Neuroscience, 33(11), 4896–4900. doi:10.1523/jneurosci.3241-12.2013
Keuken, M. C., Bazin, P. L., Crown, L., Hootsmans, J., Laufer, A., Müller-Axt, C., … Forstmann, B. U. (2014). Quantifying inter-individual anatomical variability in the
subcortex using 7T structural MRI. NeuroImage, 94, 40–46. https://doi.org/10.1016 /j.neuroimage.2014.03.032
Keuken, M. C., Van Maanen, L., Bogacz, R., Schäfer, A., Neumann, J., Turner, R., & Forstmann, B. U. (2015). The subthalamic nucleus during decision-making with multiple alternatives. Human Brain Mapping, 36(10), 4041–4052. https://doi.org /10.1002/hbm.22896
Keuken, M. C., Bazin, P. L., Backhouse, K., Beekhuizen, S., Himmer, L., Kandola, A., … Forstmann, B. U. (2017). Effects of aging on T1, T2∗ , and QSM MRI values in the subcortex. Brain Structure and Function, 222(6), 2487–2505. doi: /10.1007/s00429-016-1352-4
Keuken, M. C., Isaacs, B. R., Trampel, R., van der Zwaag, W., & Forstmann, B. U. (2018). Visualizing the Human Subcortex Using Ultra-high Field Magnetic Resonance
Imaging. Brain Topography. doi:10.1007/s10548-018-0638-7
Lenglet, C., Abosch, A., Yacoub, E., De Martino, F., Sapiro, G., & Harel, N. (2012).
Comprehensive in vivo mapping of the human basal ganglia and thalamic connectome in individuals using 7T MRI. PloS ONE, 7(1), e29153. doi:10.1371
/journal.pone.0029153
Li, W., van Tol, M. J., Li, M., Miao, W., Jiao, Y., Heinze, H. J., … Walter, M. (2014). Regional specificity of sex effects on subcortical volumes across the lifespan in healthy aging. Human Brain Mapping, 35(1), 238–247. doi:10.1002/hbm.22168 Li, W., Avram, A. V., Wu, B., Xiao, X., & Liu, C. (2014). Integrated Laplacian-based phase
unwrapping and background phase removal for quantitative susceptibility mapping.
Li, W., Wang, N., Yu, F., Han, H., Cao, W., Romero, R., … Liu, C. (2015). A method for estimating and removing streaking artifacts in quantitative susceptibility mapping.
NeuroImage, 108, 111–122. doi:10.1016/j.neuroimage.2014.12.043
Mallet, N., Micklem, B. R., Henny, P., Brown, M. T., Williams, C., Bolam, J. P., … Magill, P. J. (2012). Dichotomous Organization of the External Globus Pallidus. Neuron, 74(6), 1075–1086. doi:10.1016/j.neuron.2012.04.027
Marques, J. P., Kober, T., Krueger, G., van der Zwaag, W., Van de Moortele, P. F., & Gruetter, R. (2010). MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. NeuroImage, 49(2), 1271–1281. doi:10.1016/j.neuroimage.2009.10.002
Middleton, F. A., & Strick, P. L. (2000). Basal ganglia output and cognition: Evidence from anatomical, behavioral, and clinical studies. Brain and Cognition. doi:10.1006
/brcg.1999.1099
Odekerken, V. J. J., van Laar, T., Staal, M. J., Mosch, A., Hoffmann, C. F. E, Nijssen, P. C. G., … de Bie, R. M. A. (2013). The Lancet Neurology, 12(1), 37-44. Subthalamic nucleus versus globus pallidus bilateral deep brain stimulation for advanced Parkinson’s disease (NSTAPS study): a randomized controlled trial. doi:10.1016 /S1474-4422(12)70264-8
Plantinga, B. R., Temel, Y., Roebroeck, A., Uludağ, K., Ivanov, D., Kuijf, M. L., & ter Haar Romenij, B. M. (2014). Ultra-high field magnetic resonance imaging of the basal ganglia and related structures. Frontiers in Human Neuroscience, 8, 876. doi:10.3389 /fnhum.2014.00876
Raznahan, A., Shaw, P. W., Lerch, J. P., Clasen, L. S., Greenstein, D., Berman, R., … Giedd, J. N. (2014). Longitudinal four-dimensional mapping of subcortical anatomy in
human development. National Academy of Sciences, 111(4), 1592-1597. doi:10.1073/pnas.1316911111
Rijpkema, M., Everaerd, D., van der Pol, C., Franke, B., Tendolkar, I., & Fernández, G. (2011). Normal sexual dimorphism in the human basal ganglia. Human Brain
Mapping, 33(5), 1246-1252. doi:10.1002/hbm.21283
Smith, C. D., Chebrolu, H., Wekstein, D. R., Schmitt, F. A., & Markesberry, W. R. (2007). Age and gender effects on human brain anatomy: A voxel-based morphometric study in healthy elderly. Neurobiology of Aging, 28(7), 1075-1087. doi:10.1016
/j.neurobiolaging.2006.05.018
Sowell, E. R., Peterson, B. S., Kan, E., Woods, R. P., Yoshii, J., Bansal, R., … Toga,A. W. (2007). Sex differences in cortical thickness mapped in 176 healthy individuals
between 7 and 87 years of age. Cerebral Cortex, 17(7), 1550-1560. doi:10.1093 /cercor/bhl066
Takahashi, R., Ishii, K., Kakigi, T., & Yokoyama, K. (2011). Gender and age differences in normal adult human brain: Voxel-based morphometric study. Human Brain Mapping,
32(7), 1050-1058. doi:10.1002/hbm.21088
Zou, H. K., Warfield, S. K., Baratha, A., Tempany, C. M C., Kaus., M. R., Haker, S.J., … Kikinis, R. (2004). Statistical validation of image segmentation quality based on a spatial overlap index. Academic Radiology, 11(2), 178-189. doi:10.1016/S10766332 (03)00671-8
