Research Project 1

MSc

Brain and Cognitive

Sciences

Behavioral Neuroscience

Research Project 1

The Effect of Normal Aging on Subcortical Structures

by

Zeli Chen

12091081

October 2019

26ECs

March – October 2019

Supervisor/Examiner:

Examiner:

Dr. Anneke Alkemade

Dr. Rob de Heus

Abstract:

Reliable human brain atlases can offer benefits for contemporary functional imaging studies as well as neurosurgical procedures. Histological atlases based on post-mortem brain samples are constrained by the small subject pool of donors and therefore lacking detailed information of individuality. The applicability of probabilistic atlases instead, is constrained by low spatial resolution and limited visibility of small

subcortical structures of previous studies. Based on 7T Magnetic Resonance Imaging (MRI) scanning, this study is aimed to test the effect of age and sex on the volume, spatial properties, and MRI parameters of subcortical structures, including the Globus Pallidus Externa, Red Nucleus, Substantia Nigra, and Subthalamic Nucleus. Scanning data has been obtained from an overall of 105 subjects with a wide age range. Conjunct masks obtained from manual segmentation have been registered to the MNI space. We have identified various age-related anatomical changes, including the absolute volume, spatial properties, and quantitative MRI parameters, which suggests a need in building age-specific atlases.

Introduction:

Contemporary functional neuroimaging studies benefit from widely-applicable atlases of human brain to accurately locate cortical and subcortical structures, which take on diverse functions and add to complex large-scale connections. The quality of brain atlases is most essential for studies based on neuroimaging techniques as they put on a direct influence on how scanning data would be processed during the research and how different studies can be interpreted and integrated to get better cross-validity (Devlin and Poldrack. 2007). Besides, neurosurgeons usually refer to brain atlases for various motives, including educational purposes and to provide better clinical results. For example, the intervention of Deep Brain Stimulation in patients with Parkinson Disease consists in direct electric stimulation of subcortical structures including bilateral subthalamic nuclei (STN) and the misplacement of the electrode can lead to unwanted changes of neurological states (Walter et al. 2011).

Currently used histological atlases, like the Schaltenbrand and Wahren atlas (1977) or the Mai atlas (Mai, Majtanik and Paxinos. 2016), due to their dependency on post-mortem samples, are usually based on brains obtained from a limited number of subjects, therefore lacking information related to individual differences. Existing atlases built on structural MRI imaging, instead, are constrained by technical limitations of 1.5 tesla (T) and 3T scanning that previous studies rely on, which provides low spatial resolution and therefore not enough details, constraining in particular the visibility of small subcortical structures. Research indicates that current MRI atlases only include available mappings for 7% of known subcortical structures

(Alkemade, Keuken and Fortsmann. 2013). Besides, when considering applying those atlases to population imaging, the majority fails to include enough pre-adolescent or senior subjects to cover a fully-applicable age range (Dickie et al. 2017).

It is widely recognized that normal aging process can cause significant anatomical modifications in human brain. In general, the transition begins as early as in the early twenties of an adult when the brain shows first signs of deterioration as it shrinks weight and size (Svennerholm, Boström, & Jungbjer. 1997), which can happen in both cortical and subcortical structures (Raznahan et al. 2011, Raznahan et al. 2014), with a simultaneous loss of both white and grey matters (Giorgio et al. 2010), accompanied by an enlargement of the ventricular size (Anderton. 1997). Within the cerebral cortex, the prefrontal cortex is reported to be the most susceptible to the volume shrinkage as a result of aging (Salat et al. 2004, Raz et al. 1997). While for the subcortical area, existing studies mainly deal with larger structures, including caudate, thalamus, and hippocampus, etc. The results suggest that there are stronger age-related effects in older subjects within those regions (Goodro et al. 2012, Walhovd et al. 2005.). Besides, along with the aging process, human brain is also undergoing a gradual loss of lipid substance, indicating demyelination (Svennerholm, Boström, & Jungbjer. 1997). This effect can be reflected in MRI studies as an increase in T1 value (Callaghan et al. 2014). However, the relationship between age and MRI parameters are not consistent in previous studies, as some revealed a linear relationship (Keuken et al. 2017) while other indicates a better fitting of quadratic model (Okubo et al, 2017).

With the newly developed 7T MRI scanning, higher spatial resolution can be obtained as a result of the increased signal to noise ratio as a direct result of increased filed strength compared to prior 1.5T or 3T scanning (Novak et al. 2005), and therefore provide clearer visualizations of anatomical structures which

allow for more defined segmentation protocols and more accurate delineation of individual structures (Thomas et al. 2008). Based on 7T MRI, researchers have been studying the effect of aging on smaller subcortical structures including the red nucleus (RN), subthalamic nucleus (STN), and SN (substantia nigra) etc and proved that normal aging process can influence the volume, location as well as the image intensity of those structures (Keuken et al. 2014, Keuken et al. 2017).

Sex differences in the aging brain have also raised attention. Men and women are reported to show different diagnosis rate and clinical symptoms of neurodegenerative disorders as Parkinson’s Disease or the

Alzheimer Disease (Miller & Cronin-Golomb, 2010, Viña & Lloret, 2010). However, structural studies revealed contradictory results concerning how sex modifies the aging brain. Previous findings include that males suffer from a stressed decline in grey matter volume of subcortical structures as hippocampus (Li et al. 2014), caudate nucleus, and putamen (Kiraly et al. 2015) or stressed atrophy within the left basal ganglia (Xu et al. 2000). While there are also reports that indicate faster pace of volume decline in women within the basal ganglia area (Li et al. 2014). Also, evidence also exists that suggests no sex difference for grey matter loss in those regions (Ge et al. 2002, Sullivan et al. 2004).

In order to get clear views of how age and sex modify the anatomical features of the brain, and how those features would influence the validity of current-existing atlases when applied to senior subjects, as a part of a larger project, this study is based on 7T MRI scanning data collected from 110 subjects with a wide age range. Subcortical structures, including the globus pallidus externa (GPe), RN, SN, and STN, are targeted in this study, focusing on potential changes related to the volume, spatial properties and quantitative MRI parameters of those regions.

Methods

Subjects:

The study is based on prior work of an on-going project led by the Integrative Model-based Cognitive Neuroscience lab (IMCN) at the University of Amsterdam. Complete datasets have been previously collected from an overall of 105 subjects, aging from 19 to 80 (mean = 42.3, SD = 19.3, 60 females) and restored in the AHEAD database. Due to missing data of STN masks, only data from 103 subjects (mean = 41.8, SD = 19.2, 59 females) were used for later analysis of STN volume and the effect of location. All participants had corrected to normal vision. They have given written consent prior to the scanning and none of them had prior reports of neurological, physical or psychological disorders as assessed by self-reports.

Scanning and Image Processing:

All the subjects underwent a single MP2RAGEME (Caan et al. 2018) sequence brain scanning using a Philips Achieva 7T MRI scanner with a 32-channel Nova head coil (NOVA Medical Inc., Wilmington MA) at the Spinoza Center for Neuroimaging. The MP2RAGEME sequence is a combination of the multi-echo technology and the second gradient‐echo image of the MP2RAGE sequence (Marques et al., 2010), which allows for the possibility to obtain T1, T2* and QSM information with one single undertake. The scanning lasts a duration of 16.5 minutes (TR = 6.72s; TE1/TE2/TE3/TE4 = 3/11.5/20/28.5ms; TI1/TI2 =

670/3855ms; flip angle = 7°/6°; bandwidth = 405 Hz/Px). A whole brian scan (voxel size = 0.64 x 0.64 x 0.70mm) as well as a slab (voxel size = 0.5 x 0.5 x 0.5mm) was obtained. The details of preprocessing can be found in the paper of Caan and colleagues (2018). In short, the T1 maps were computed by an iterative 2d interpolation of the T1 and B1+ look up tables to correct for residual transmit B1+ inhomogeneities (Marques & Gruetter, 2013). The T2* weighted maps were obtained by using a single exponential fit. The QSMs were obtained after applying the Laplacian operator to unwrap the measured phase images and remove the background phase information using the STI suite (Li et al. 2013). The sequence allows for simultaneous acquisition of all three images from one single Fourier space, therefore, structures parcellated on different contrasts can be directly put into analysis without the necessity of cross-reference and

coregistration (Caan et al. 2018).

Manual Segmentation:

The segmentation procedure was carried out by several different raters using FSLeyes

(https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSLeyes). Due to the number of subjects that participated in the study

and the innate workload of segmentation work, the two raters whose masks are used to build the conjunct mask are not consistent for each individual structure. Each conjunct mask of the structures used for further analysis is constructed based on the masks of two independent raters and only voxels marked by both raters are included. To guarantee the consistency during segmentation, all the structures are parcellated based on protocols previously developed in the lab combining anatomical studies, MRI scanning and existing human atlases. The parcellation of the structures is conducted based on QSM volumes.

After the segmentation, in order to test how much the two raters agree with each other, we obtained the Dice coefficient (Dice 1945) as a reference, which is a statistical measure of the spatial overlap accuracy

based on the volume of the conjunct mask and the masks built by each rater (Zou et al. 2014). The Dice coefficient is calculated with the follow equation:

Dice = 2 x Volume conjunct / (Volumerater1 + Volumerater2)

The volume of each structure is then extracted from the original individual space of the MRI data and used for later analysis.

All the manual segmentation procedure for the aforementioned structures have been done before hand. The author of this report performed manual delineations on the amygdala. Given that a second rater for this structure is not yet available, further analyses of these data were not pursued within the context of the current studies.

Data Registration and Acquisition of MRI parameters:

The registration was completed using the Nighres package (https://github.com/nighres). The scans were skull-stripped and reconstructed following a MP2RAGE segmentation pipeline as described in Caan et al (2018). From the segmentation median T1, T2* and QSM values were extracted from all four structures using the conjunct mask built on manual masks. The MP2RAGEME slab was previously registered to the whole brain volume and later non-linearly transformed to the MNI152 (http://nist.mni.mcgill.ca/?p=904) template. The conjunct masks of each structure were added together and fitted into the MNI template to create a probabilistic atlas of the structure.

Aging effect on the location

To test the effect of age on the location of the structures, we first obtained the coordinates (in format [x, y, z]) of the Center of Mass (CoM) for each separate structure from the conjunct masks transformed into the MNI space. The coordinates are recorded based on the relative position compared to the whole-brain image in FSLeyes by counting the voxels that are marked in the mask. By averaging the coordinates of the CoM from all four structures, we define a further center of the system which consists of all aforementioned subcortical structures, which is represented by the following equation (See Figure 1 for the schematic illustration of the CoS):

Center of System = (∑Xi/i, ∑Yi/i, ∑Zi/i)

To reduce the dimension of the coordinates, we conducted a Principle Component Analysis (PCA) per structure on the difference between x,y,z coordinates of CoM and the CoS using the Scikit-Learn package in Python. The input matrix is expressed as follows (take STN as example):

[XSTNsubject1 - XCoSsubject1, YSTNsubject1 - XCoSsubject1, ZSTNsubject1 - XCoSsubject1 XSTNsubject2 - XCoSsubject2, YSTNsubject2 - XCoSsubject2, ZSTNsubject2 - XCoSsubject2 ....

XSTNsubject - XCoSsubject, YSTNsubject - XCoSsubject, ZSTNsubject - XCoSsubject]

The loading scores of the principle component extracted of each structure, were multiplied with the

the system on the direction of the principle component.

Previous research indicates that the volume of the third ventricle increases with age (Keuken et al. 2017), which will possibly lead to the enlargement of the distance between the subcortical structures from the left and right hemisphere of the brain. In order to check for the phenomenon that can cause shifts of the center of the system itself, as its coordinates were based on the spatial properties of the structures, we calculated the distance between the left and right sides of each structure, as well as the two CoS of both sides.

Distance = ((Xleft – Xright)^2 + (Yleft – Yright)^2 + (Zleft – Zright)^2) ^ 0.5

The two distances were correlated with age to see if any significant effect exists.

Figure 1. Schematic Illustration of the CoS.

Statistical Analysis:

The statistical tests are conducted with SPSS 25 (SPSS Inc. 2017) and R studio (https://rstudio.com/). A general linear model (GLM) was adopted to test the effect of age and gender on the volume, location and MRI parameters of the structures. The test was done by setting age and sex as the main effects, the age x sex interaction as a third term to the model. A Pearson’s correlation test, as well as a paired t test was used to determine if there is lateralization effect between the left and right sides of each structure. Based on

contradictory findings from previous literature, we also included a linear and quadratic regression model for the MRI parameters to see which is a better fit. A nested model F test was used to test the model fitting. To avoid a too stringent correction and balance the power of the analysis, a Benjamini and Hochberg (1995) correction was adopted to correct for the false discovery rate by using an online calculator

(https://www.sdmproject.com/utilities/?show=FDR). All p values reported are after correction unless stated otherwise.

Results

Demographics:

The distribution of age in different sex groups is as shown in Figure 2. The student t test revealed no significant difference between the means of age between male (mean = 44.11, SD = 19.336) and female subjects (mean = 41.00, SD = 19.433); t(103) = -.814, p = .719.

Figure 2: The distribution of age in male and female subjects.

Inter-rater reliability:

The inter-rater agreement, represented by the Dice coefficient is highest on RN while raters tend to disagree more on STN. See Table 1 for mean values and SD of the Dice Coefficient per structure.

There is main effect of age on average Dice coefficient in SN (F(1,101) = 4.88, p = 0.046) and STN (F(1,99) = 6.42, p = 0.024), resulting from negative correlation between age and Dice. No sex effect or age x sex interaction was found.

Lateralization:

There is strong positive correlation between the volumes of both sides in GPe, SN and RN (GPe: r(105) = .70, p < .001, RN: r(105) = .79, p < .001, SN: r(105) = .84, p < .001). Statistical tests also indicate that the SN is slightly bigger on the right side (SN: t(104) = -3.19, p = .008). See Table 1 for separate and average volumes of the structures.

Aging effect on volume:

Results suggest that aging leads to decreased average individual volumes of GPe, RN and SN (GPe: F(1,101) = 18.20, p < .001; RN: F(1, 101) = 16.66, p < .001; SN: F(1,101) = 6.88, p = 0.019), no age effect was found in STN. No main effect of sex or age x sex interaction was supported.

Because of the lateralization effect of SN, we also conducted a separate analysis to test the volume of the left and right SN, which results in similar negative correlation between age and volume. We also observed an effect of age x sex interaction in the left SN; F(1,101) = 4,41, p (before correction) = 0.038, which, however, did not survive multiple comparison corrections.

Aging effect on location:

Results revealed a main effect of age on the location of RN, SN and STN compared to the center of the four structures. See Table 2 for the p value and F value with degree of freedom of each structure.

The distance between the left and right side of all structures increase with age (F(1,101) = 52.12, p < .001; RN: F(1,101) = 68.64, p < .001; SN: F(1,101) = 43.92, p < .001; STN: F(1, 99) = 36.92, p < .001), and this effect persists for the CoS (F(1, 99) = 11.11, t(101) = .001) .

Aging effect on T1 value:

The GLM suggests a main effect of age on T1 value in the SN (F(1,101) = 9.45, p = .007). No other effects were found. Applying linear and quadratic curves separately between age and T1 value suggests a better fitting of the quadratic model in all four structures (GPe: F(1,103) = 39.49, p < .001; RN: F(1,103) = 24.92, p < .001; SN: F(1, 103) = 10.53, p = .002; STN: F(1, 101) = 7.85, P =.006). See Table 3 for average value and standard deviation of the quantitative MRI parameters and Table 4 for the statistical parameters of the two models. Figure 3 illustrates the fitting lines of the linear and quadratic model.

Aging effect on T2* value:

The T2* value of the left GPe in one subject results negative due to calcification of the brain and therefore the GPe data of the subject is disregarded as an outliner. Evidence suggests negative correlation between the T2* value and age in RN, SN and STN (RN: F(1,101) = 52.66, p < .001; SN: F(1,101) = 18.31; p < .001; STN: F(1,99) = 15.33; p < .001), and a weak negative correlation in GPe (F(1,100) = 4.32, p (before correction) = .04) which disappears after correction for multiple comparisons. No sex or interaction effect was observed. The F test favors a quadratic model in GPe, RN and (GPe: F(1,103) = 9.51, p = .003; RN: F(1,103) = 31.45, p < .001; SN: F(1, 103) = 13.72, p < .001), while no significant difference was found for STN. See figure 4 for the scatterplot and fitting lines.

Aging effect on QSM value:

No main effect of sex or interaction was detected. Evidence suggests a positive correlation between QSM value and age in GPe and RN (GPe: F(1, 101) = 5.09, p = .043; RN: F(1, 101) = 15.70, p < .001). There is also a marginal effect of age in STN (F(1, 99) = 4.57, p (before correction) = .035), which doesn’t survive the multiple comparison.

Quadratic regression was tested as significant in GPe, RN and STN. The F test prefers a quadratic model in RN (F(1, 103) = 19.42, p < .001), while no significant difference was found between the two models in GPe and STN.

Discussion

In this study, we investigated the effect of normal aging process on the anatomical features of the brain, focusing on subcortical structures including GPe, SN, RN and STN. By using ultra high resolution 7T MRI scanning, we were able to better localize and identify smaller structures and develop effective protocols for manual segmentation procedure, which enables a more accurate calculation of relative parameters in terms of volume and spatial properties (Thomas et al. 2008). The subcortical structures tested in the research, involving the GPe, SN and STN, are main components of the basal ganglia, which is responsible for motor learning and executive functions (Lanciego, Luquin & Obeso. 2012). The RN instead, is an important relay center for motor information that situates at the level of the SN (Mai and Paxinos. 2012). To study these structures provides a clearer perspective to better understand the age-related changes of the human motor system and can have clinical significance especially in helping locating the STN. Besides, all four structures are reported to show high iron concentration (Haacke et al. 2005, Dorment et al. 2004), which makes them clearly visible on susceptibility maps (Schäfer, et al. 2011). This helps the early development of

segmentation protocols during the research.

The inter-rater reliability of the masks, which is represented by the Dice Coefficient (See Table 1), is generally comparable to the findings previously reported (Keuken et al. 2014, Keuken et al. 2017), which suggests a good to excellent agreement between the raters and the validity of the segmentation protocols. However, the average volumes of SN and STN are notably larger. The divergence may be caused by

different contrasts based on which the segmentation protocols are developed. Our study using QSM contrast revealed comparable STN volume as reported in prior research also referring to segmentation protocol based on QSM image (Alkemade et al. 2017). Because of the lack of literature in the field, further replication would be needed to confirm the scale of the volume.

We have discovered that normal aging has a general influence on various anatomical features of the

subcortical structures during the process. First, results suggest decreased volume of GPe, RN and SN across age, which is somewhat contradictory to the previous study by Keuken and colleagues (2017) but in line with various MRI studies supporting brain grey matter loss across age (Resnick et al., 2003, Raz et al., 2005). In terms with spatial properties, we have built a common reference point by averaging the

coordinates of the mass center of each individual structure and identified that relative positions of SN, RN and STN to the center tend to shift in one consistent direction across age. Together with the effect we also observed a dilation of the distance between the left and right sides in all four structures, which is also reflected on the chosen reference point. The phenomenon can be caused by the enlargement of the third ventricle, as has been suggested by previous work (Fjell and Walhovd 2010). To eliminate further variants, future research may consider add the volume of the third ventricle as a covariant for the analysis.

By adopting a common reference point, potential risks are that some internal position shifts can be overlooked between the structures. And our findings actually revealed a trend of occurrence that the four structures tend to change their relative positions related to each other during the process. Therefore, even if statistical tests do not support significant variance of the GPe compared to the center, it still may happen that its absolute position actually shifts in the same direction with the reference point during the aging process. Besides, by adding the distance between the left and right hemisphere, we also proved that age can

affect the position of those structures relative to anterior commissure - posterior commissure (AC-PC) line of the brain. Both effects can be of important hints for DBS protocols in localizing the STN, as a common method to determine the coordinates for electrode placement consists in referring to mid-commissure point and the neighboring RN as landmarks (Bejjani et al. 2000, Rabie, Metman & Slavin. 2016).

The quantitative MRI parameters of the structures are affected by the myelination degree of the axons and the iron concentration level. In general, increased T1 and T2* values can stand for demyelination and low iron deposit (Ogg & Steen, 1998, Okubo et al. 2017), while increased quantitative magnetic susceptibility is linked to an increase in myelin and iron (Deistung et al. 2013). Our results suggest that the quadratic curve is a better fitting model for describing the relationship between T1 value and age in all four structures, while the difference between the two models is more trivial concerning the T2* and QSM values. Mid-age adults tend to score the lowest in T1 and T2* values and have relatively high QSM value (See Figure 3-5). The finding is consistent with previous literature (Okubo et al. 2017) suggesting a better fitting quadratic curve of T1 value with age in SN and GP but contradictory to the findings of Keuken and colleagues (2017). In anatomical studies, T1 weighted images are widely used for developing automated segmentation protocols (Beller et al. 2018). New algorithms based on deep learning are also being developed combining images obtained from different contrasts to get better segmentation results (Akkus et al. 2017, Feng et al. 2017). A better understanding of the age-related changes would help reduce the age-related bias during the automated segmentation process (Keuken et al. 2017).

We have also identified moderate negative correlation between age and Dice Coefficient in SN and STN. As a relatively small structure, the STN can be more susceptible than others to the subjectivity of the raters and can be affected by the unclear borders between these two structures. Therefore, the results should be examined with caution and combined with further shape analysis. Also, both the analysis for volume and location seem to suggest a lateralization effect in SN. Although those effects will most likely disappear with a less conservative method of correction for multiple comparison.

Surprisingly, we didn’t find main effects of sex or age x sex interaction in any of the parameters measured. The results do not support our initial hypothesis that males and females may age with different speed and are in line with prior research suggesting no sex difference during the aging process in grey matter volume (Jäncke et al. 2014). Several points should be considered when interpreting the results. First, males and females have different brain volumes, and the volumes of the structures haven’t been corrected for the inter-cranial volume in our research; second, although we include subjects with a wide age range, we have included fewer senior subjects compared to younger ones, while the sexual dimorphism within the brain is more likely to happen in a relative late age (Coupé et al. 2017); third, male and female subjects can follow different trajectories better fitting with higher order models which have not been tested in this study.

Therefore, we suggest predefined age groups to better study the differences caused by sex in senior subjects. Apart from aforementioned limitations, due to the difficulties in keeping track of the tested subjects, it is not possible to conduct large-scale longitudinal studies on age-related research. However, a cross-sectional design of the study would inevitably lead to more uncontrollable variance although the incongruence of age-related changes can also happen in longitudinal-designed studies (Coupé et al. 2017). By adding the

diminished statistical power. Also, although manual segmentation is generally viewed as a standard criterion compared to automated procedures, the process is not free of flaws. In this study, the contrasts of the images used for delineation are changing across age, which can lead to inconsistent visibility of the structures.

Overall, we have identified various age – related anatomical changes in different structures, including the absolute volume, spatial properties and quantitative MRI parameters, which suggests a need in building age - specific atlases of the structures, especially for elder subjects.

Reference

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

Anderton BH (1997): Changes in the ageing brain in health and disease. Philos Trans R Soc Lond B Biol Sci 352:1781–1792.

Akkus, Z., Galimzianova, A., Hoogi, A., Rubin, D. L., & Erickson, B. J. (2017). Deep Learning for Brain MRI Segmentation: State of the Art and Future Directions. Journal of Digital Imaging, 30(4), 449–459. doi:10.1007/s10278-017-9983-4

Beller, E., Keeser, D., Wehn, A., Malchow, B., Karali, T., Schmitt, A., … Stoecklein, S. (2018). T1-MPRAGE and T2-FLAIR segmentation of cortical and subcortical brain regions—an MRI evaluation study. Neuroradiology. doi:10.1007/s00234-018-2121-2

Bejjani, B.-P., Dormont, D., Pidoux, B., Yelnik, J., Damier, P., Arnulf, I., … Cornu, P. (2000). Bilateral subthalamic stimulation for Parkinsons disease by using three-dimensional stereotactic magnetic resonance imaging and electrophysiological guidance. Journal of Neurosurgery, 92(4), 615–625. doi: 10.3171/jns.2000.92.4.0615

Benjamini, Yoav; Hochberg, Yosef (1995). "Controlling the false discovery rate: a practical and powerful approach to multiple testing" (PDF). Journal of the Royal Statistical Society, Series B. 57 (1): 289–300. MR 1325392.

Caan, M. W. A., Bazin, P. L., Marques, J. P., Hollander, G., Dumoulin, S. O., & Zwaag, W. (2018). MP2RAGEME: T1, T2*, and QSM mapping in one sequence at 7 tesla. Human Brain Mapping, 40(6), 1786–1798. doi: 10.1002/hbm.24490

Callaghan MF, Freund P, Draganski B et al (2014) Widespread age- related differences in the human brain microstructure revealed by quantitative magnetic resonance imaging. Neurobiol Aging NBA 35:1862–1872. doi: 10.1016/j.neurobiolaging.2014.02.008

Coupé, P., Catheline, G., Lanuza, E., & Manjón, J. V. (2017). Towards a unified analysis of brain maturation and aging across the entire lifespan: A MRI analysis. Human Brain Mapping, 38(11), 5501–5518.

doi:10.1002/hbm.23743

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

Devlin, J. T., & Poldrack, R. A. (2007). In praise of tedious anatomy. NeuroImage, 37(4), 1033–1041. doi: 10.1016/j.neuroimage.2006.09.055

Dice, L. R. (1945). Measures of the Amount of Ecologic Association Between Species. Ecology, 26(3), 297–302. doi: 10.2307/1932409

Dickie, D. A., Shenkin, S. D., Anblagan, D., Lee, J., Cabez, M. B., Rodriguez, D., … Wardlaw, J. M. (2017). Whole Brain Magnetic Resonance Image Atlases: A Systematic Review of Existing Atlases and Caveats for Use in Population Imaging. Frontiers in Neuroinformatics, 11. doi: 10.3389/fninf.2017.00001

Dormont D, Ricciardi KG, Tande D, Parain K, Menuel C, Galanaud D, Navarro S, Cornu P, Agid Y, Yelnik J ( 2004): Is the subthalamic nucleus hypointense on T2‐weighted images? A correlation study using MR imaging and stereotactic atlas data. AJNR Am J Neuroradiol 25: 1516–1523.

Goodro, M., Sameti, M., Patenaude, B., & Fein, G. (2012). Age effect on subcortical structures in healthy adults. Psychiatry Research: Neuroimaging, 203(1), 38–45. doi: 10.1016/j.pscychresns.2011.09.014

Feng, X., Deistung, A., Dwyer, M. G., Hagemeier, J., Polak, P., Lebenberg, J., … Schweser, F. (2017). An improved FSL-FIRST pipeline for subcortical gray matter segmentation to study abnormal brain anatomy using

quantitative susceptibility mapping (QSM). Magnetic Resonance Imaging, 39, 110–122. doi:10.1016/j.mri.2017.02.002

Fjell, A. M., & Walhovd, K. B. (2010). Structural Brain Changes in Aging: Courses, Causes and Cognitive Consequences. Reviews in the Neurosciences, 21(3). doi: 10.1515/revneuro.2010.21.3.187

Haacke, E. M., Cheng, N. Y., House, M. J., Liu, Q., Neelavalli, J., Ogg, R. J., … Obenaus, A. (2005). Imaging iron stores in the brain using magnetic resonance imaging. Magnetic Resonance Imaging, 23(1), 1–25. doi: 10.1016/j.mri.2004.10.001

Jäncke, L., Mérillat, S., Liem, F., & Hänggi, J. (2014). Brain size, sex, and the aging brain. Human Brain Mapping, 36(1), 150–169. doi:10.1002/hbm.22619

Király, A., Szabó, N., Tóth, E., Csete, G., Faragó, P., Kocsis, K., . . . Kincses, Z. T. (2015). Male brain ages faster: The age and gender dependence of subcortical volumes. Brain Imaging and Behavior,10(3), 901-910. doi:10.1007/s11682-015-9468-3

Lanciego JL, Luquin N, Obeso JA. Functional Neuroanatomy of the Basal Ganglia. Cold Spring Harb Perspect Med. (2012) 2: a009621–a009621. 10.1101/cshperspect.a009621

Li, W., Avram, A. V., Wu, B., Xiao, X., & Liu, C. (2013). Integrated Laplacian-based phase unwrapping and background phase removal for quantitative susceptibility mapping. NMR in Biomedicine, 27(2), 219–227. doi: 10.1002/nbm.3056

Li, W., Tol, M.-J. V., Li, M., Miao, W., Jiao, Y., Heinze, H.-J., … Walter, M. (2012). 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

Mai Jürgen K., & Paxinos, G. (2012). The human nervous system. Amsterdam: Elsevier.

Mai, J. K., Majtanik, M. and Paxinos, G. (2016) Atlas of the Human Brain.

Marques, J. P., & Gruetter, R. (2013). New developments and applications of the MP2RAGE sequence - Focusing the contrast and high spatial resolution R1 mapping. PLoS One, 8, e69294.

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, 1271–1281.

Miller, I. N., & Cronin-Golomb, A. (2010). Gender differences in Parkinsons disease: Clinical characteristics and cognition. Movement Disorders, 25(16), 2695–2703. doi: 10.1002/mds.23388

Ogg, R. J., & Steen, R. G. (1998). Age-related changes in BrainT1 are correlated with iron concentration. Magnetic Resonance in Medicine, 40(5), 749–753. doi:10.1002/mrm.1910400516

Okubo, G., Okada, T., Yamamoto, A., Fushimi, Y., Okada, T., Murata, K., & Togashi, K. (2017). Relationship between aging and T 1 relaxation time in deep gray matter: A voxel-based analysis. Journal of Magnetic Resonance Imaging, 46(3), 724–731. doi: 10.1002/jmri.25590

Rabie, A., Metman, L. V., & Slavin, K. (2016). Using “Functional” Target Coordinates of the Subthalamic Nucleus to Assess the Indirect and Direct Methods of the Preoperative Planning: Do the Anatomical and Functional Targets Coincide? Brain Sciences, 6(4), 65. doi: 10.3390/brainsci6040065

Raz, N., Gunning FM, Head D, Dupuis JH, McQuain J, Briggs SD, Loken WJ, Thornton AE, Acker JD. (1997). Selective aging of the human cerebral cortex observed in vivo: differential vulnerability of the prefrontal gray matter. Cerebral Cortex, 7(3), 268–282. doi: 10.1093/cercor/7.3.268

Raz N, Lindenberger U, Rodrigue KM, Kennedy KM, Head D, Williamson A, Dahle C, Gerstorf D, Acker JD. Regional brain changes in aging healthy adults: general trends, individual differences and modifiers. Cereb. Cortex. 2005;15:1676–1689.

Raznahan, A., Shaw, P., Lalonde, F., Stockman, M., Wallace, G. L., Greenstein, D., ... Giedd, J. N. (2011). How does your cortex grow?. The Journal of neuroscience: the official journal of the Society for Neuroscience, 31(19), 7174–7177. doi:10.1523/JNEUROSCI.0054-11.2011

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. Proceedings of the National Academy of Sciences,111(4), 1592-1597. doi:10.1073/pnas.1316911111

Resnick SM, Pham DL, Kraut MA, Zonderman AB, Davatzikos C. Longitudinal magnetic resonance imaging studies of older adults: a shrinking brain. J. Neurosci. 2003;23:3295–3301.

Salat, D. H., Buckner RL, Snyder AZ, Greve DN, Desikan RS, Busa E, Morris JC, Dale AM, Fischl B. (2004). Thinning of the Cerebral Cortex in Aging. Cerebral Cortex, 14(7), 721–730. doi: 10.1093/cercor/bhh032

Schäfer, A., Forstmann, B. U., Neumann, J., Wharton, S., Mietke, A., Bowtell, R., & Turner, R. (2011). Direct visualization of the subthalamic nucleus and its iron distribution using high-resolution susceptibility mapping. Human Brain Mapping, 33(12), 2831–2842. doi: 10.1002/hbm.21404

Schaltenbrand, G., Hassler, R., & Wahren, W. (2005). Atlas for stereotaxy of the human brain: with an accompanying guide. Stuttgart: Thieme.

Sullivan, E. V., Rosenbloom, M., Serventi, K. L., & Pfefferbaum, A. (2004). Effects of age and sex on volumes of the thalamus, pons, and cortex. Neurobiology of Aging, 25(2), 185–192.

Svennerholm, L., Boström, K., & Jungbjer, B. (1997). Changes in weight and compositions of major membrane components of human brain during the span of adult human life of Swedes. Acta Neuropathologica, 94(4), 345– 352. doi: 10.1007/s004010050717

Xu J, Kobayashi S, Yamaguchi S, et al. Gender effects on age-related changes in brain structure. AJNR Am J Neuroradiol 2000;21:112–18

Viña, J., & Lloret, A. (2010). Why Women Have More Alzheimers Disease Than Men: Gender and Mitochondrial Toxicity of Amyloid-β Peptide. Journal of Alzheimers Disease, 20(s2). doi: 10.3233/jad-2010-100501

Walter, U., Kirsch, M., Wittstock, M., Müller, J.-U., Benecke, R., & Wolters, A. (2011). Transcranial Sonographic Localization of Deep Brain Stimulation Electrodes Is Safe, Reliable and Predicts Clinical Outcome. Ultrasound in Medicine & Biology, 37(9), 1382–1391. doi: 10.1016/j.ultrasmedbio.2011.05.017

Zou, K. H., Warfield, S. K., Bharatha, A., Tempany, C. M., Kaus, M. R., Haker, S. J., … Kikinis, R. (2004). Statistical validation of image segmentation quality based on a spatial overlap index1. Academic Radiology, 11(2), 178– 189. doi: 10.1016/s1076-6332(03)00671-8

Appendix:

The manual segmentation protocol of amygdala was developed previously and is executed on T1 weighted images, on which the amygdala appears as a dark grey region. The author of the report performed manual delineations of bilateral amygdala on the T1 image of 105 subjects. These data require delineations from a second rater before they can be interpreted in a meaningful way. Therefore further analyses were not pursued within the context of this internship. The author would like to note the following for future

analyses: In the axial view of mid dorsal-ventral section, the anterior amygdaloid area is separated from the lateral nucleus by a clear myelinated white fiber band (Tyszka & Pauli. 2016). Previous study suggests that with normal aging process, there is decreased overall contrast between the grey/white matter (Salat et al. 2009). Our research instead revealed a quadratic curve of T1 contrast and age. When conducting manual segmentation, this white bundle can be wrongly viewed as a border that defining anterior end of amygdala and therefore cause an underestimation of amygdala volume especially within younger subjects.

Figure 1: An axial view of amygala on T1(left) and T2(right) weighted MRI scanning image

(Tyszka & Pauli. 2016 ).

Reference

:

Salat, D. H., Lee, S. Y., van der Kouwe, A. J., Greve, D. N., Fischl, B., & Rosas, H. D. (2009). Age-associated alterations in cortical gray and white matter signal intensity and gray to white matter contrast. NeuroImage, 48(1), 21–28. doi:10.1016/j.neuroimage.2009.06.074

Tyszka, J. M., & Pauli, W. M. (2016). In vivo delineation of subdivisions of the human amygdaloid complex in a high-resolution group template. Human Brain Mapping, 37(11), 3979–3998. doi:10.1002/hbm.23289