Development of a segmentation protocol for the dorsal thalamus using
MP2RAGEME images at 7 Tesla
Simon. R. Poortman
Integrative Model-based Cognitive Neuroscience Research Unit, University of Amsterdam, Amsterdam, The Netherlands
R E S E A R C H P R O J E C T I N F O
Credits: 26 EC
Starting date: 7 February 2017 Final date: 16 June 2017
Student name & ID: Simon Poortman, 10193820
Supervisor: dr. M. J. (Martijn) Mulder UvA representative: dr. J. M. (Anneke) Alkemade
Co-assessor: dr. J. M. (Anneke) Alkemade
Integrative Model-based Cognitive Neuroscience Research Unit MSc in Brain & Cognitive Sciences, University of Amsterdam, Cognitive Neuroscience
A B S T R A C T
In vivo visualization of subcortical structures with magnetic resonance
imaging (MRI) has proven to be a challenging endeavour due to their architectural intricacy and deep position within the brain, hindering functional MRI research. The thalamus, a relay station of the brain, lies innermost in the subcortex and shares a wide range of connections with other brain areas, making it a vital structure for cognitive functioning and behaviour. State-of-the-art MRI techniques allow us to overcome the problem of difficult visualization of the thalamus and other subcortical structures. To map these structures, manual segmentation of each individual structure is required. In this report, a validated protocol is presented for manual segmentation of the thalamus using ultra-high field MRI. The scans of six healthy control subjects were randomly selected from a database acquired from the Integrative Model-based Cognitive Neuroscience Research Unit (IMCN) at the University of Amsterdam, as part of an ongoing study aiming to create a probabilistic atlas of the human subcortex. MRI scans were performed using a 7T Philips MR scanner, with a 32-channel head coil. High resolution images were acquired using a MP2RAGEME pulse sequence, with resulting voxel size 0.64 x 0.64 x 0.7 mm³. Manual segmentation of the left and right thalami was performed on 43 consecutive coronal sections, using FslView 4.0.1 software. In addition, sagittal and axial views were included where deemed necessary. Thalamic volumes were measured and reliability of image analysis was assessed by calculating the intra-rater correlations between the first and second segmentation of each thalamus (left and right, resulting in a total of 10 thalami). Intra-rater correlations of 0.91 and 0.94 indicate that this protocol serves as a reliable tool for mapping of the thalamus. With the mapping of more subcortical structures and registration of these maps into standard space a probabilistic atlas can be created, greatly facilitating fMRI research on the subcortex.
KEY WORDS: thalamus, 7T, MP2RAGEME, manual segmentation, subcortex, atlas
Introduction
Imaging the human subcortex
The human subcortex is a complex network, comprising approximately 455 individual structures that make up roughly a quarter of the entire human brain volume (Alkemade et al., 2013). Only a mere seven percent of these structures is depicted in the standard MRI-atlases currently available (Alkemade et al., 2013). The lack of in vivo MRI research on the subcortex can be explained by its architectural intricacy and its location deep inside the brain. While in vivo imaging techniques are improving, the acquired anatomical detail
remains limited (Osechinskiy & Kruggel, 2009; Annese, 2012). It is their small size and high spatial densitythat cause subcortical nuclei to be difficult to visualize with MRI at field strengths of 3 Tesla (3T) and below (de Hollander et al, 2015). A standard anatomical scan obtained with a 3T system takes approximately six minutes, providing an isotropic resolution of one millimetre (Alkemade et al., 2013). Although acquiring an image with this detail in such a short amount of time is impressive, it does not suffice for the visualization of individual neurons, which can be attained using histological approaches (Duyn, 2012). Increasing field strength of MRI scanners enhances the spatial resolution and thus the signal-to-noise ratio (SNR) (Rutt & Lee, 1996; Maubon et al., 1999). Using ultra-high field 7T
MRI helps to visualize smaller regions such as subcortical areas (Schafer et al., 2012). Yet, even when comparing these scans to histologically stained brain sections of that same area, the difference in detail is immediately apparent (Bridge & Clare, 2006).
Taking abovementioned limitations into consideration, it is unlikely that in vivo MRI will reach the level of detail that is obtained using histological approaches. However, the digital utility of MRI-atlases remains a highly valuable factor. MRI can be used in a standardized 3D coordinate frame for data analysis and reporting of findings from fMRI research. This enables the combination of data from multiple subjects such that a probabilistic atlas can be created. These digital, population-based brain atlases allow for quantitative characterization of both anatomical variability across specific groups and the relationship of these anatomical measurements with behavioural performance. Moreover, data from a probabilistic atlas can be subdivided into groups based on specific criteria such as disease, age, sex and other demographics, providing a brain template representative of specific subject samples in fMRI experiments. While a number of human brain atlases exist, these are far from complete. Many of these atlases are based on a single or only a few brains (e.g., the Talairach & Tournoux atlas, 1998), constructed using MRI systems of 3T or below and, as mentioned earlier, lacking small brain structures, especially subcortical structures (Evans et al., 2012). As fMRI signals are often indistinct and multi-interpretable – especially so in the subcortex –, using these templates does not suffice for the precise allocation of this activity to specific brain areas. Therefore, the aim of the current project is to develop a brain atlas that accounts for these limitations.
This project is operated by the Integrative Model-based Cognitive Neuroscience (IMCN) Research Unit, a group of expert neuroscientists and -anatomists dedicated to gaining a better mechanistic understanding of cognitive processes (e.g., decision-making) in the healthy and diseased brain. With access to a state-of-the-art 7T MRI system and newly developed scanning sequences, the unit can focus on the visualization of smaller and deeply located brain structures. Furthermore, unlike existing histological atlases, the atlas of this project will comprise at least 60 individual brains, of which an average will be made to account for variability in structure location. In addition, because of this large-scale database, variance in anatomical metrics as a function of age or sex can be captured. Lastly, this will be an atlas of the scarcely studied subcortex, which will greatly facilitate fMRI research on subcortical mechanisms.
To map the subcortex in vivo, a region-of-interest approach will be employed, involving the manual segmentation of individual subcortical structures. This means that a trained rater manually segments individual structures in the
MRI scan of each single subject (Andreasen et al., 1990; Flaum et al., 1995; Portas et al., 1998; Gur et al., 1998; Spinks et al., 2002). Considering the complexity of the subcortex and its sheer number of substructures, manual segmentation of the subcortex is a laborious and time-consuming endeavour. Hence, to divide the workload in a structural manner, each researcher within the IMCN project has been assigned one or two subcortical structures to segment and for which to set up a protocol. This report specifically focuses on one of the largest subcortical structures, the thalamus. The thalamus is a large structure with a variety of functions. Each of its small nuclei has one or multiple roles in the processing of all kinds of information, from sensory information to higher cognitive functions. To better investigate the role of the human thalamus and how its functions relate to its connectivity with other regions, a detailed map of the thalamus and its substructures is crucial.
Other approaches than manual segmentation, employing automated components, have been used to examine the thalamus in MRI (e.g., image averaging, edge-finding) (Andreasen et al., 1994; Buchsbaum et al., 1996). However, whereas automated segmentation drastically increases the speed of the process and could be useful for dealing with large samples, the used tools are still susceptible to systematic bias and less suitable for accurate boundary definition (Visser et al., 2016). Furthermore, accounting for variability between individual brains of patients is challenging when using these tools (Spinks et al., 2002).Lastly, the process of automated segmentation still requires validation as manual segmentation does. Thus, albeit laborious, the manual approach seems to be better option for segmentation.
To account for inter-rater variability, manual segmentations of a region of interest would have to be performed using the same rules and training for each rater, especially for regions that are difficult to delineate. Because of the limited contrast between specific structures and neighbouring areas, segmentation may be dependent on arbitrary landmarks. For example, the exact border between a grey matter structure and white matter may be unclear when there are white matter fibres running through the grey matter structure as well, due to partial volume effects. Furthermore, the relative position of a certain nucleus to its neighbouring structures may vary per individual brain. While it is inevitable that even the most experienced anatomists will conduct different delineations, detailed protocols containing step-by-step segmentation instructions keep this inter-rater inconsistency at a minimum.
Anatomy and landmarks of the thalamus
Deriving from the embryonic vertebrate forebrain, the diencephalon consists of four distinct structures: the subthalamus, the
hypothalamus, the epithalamus and the thalamus. The latter is a paired, symmetrical structure joined at the midline, located near the centre of the brain. In humans, each of the thalami is an oval-shaped structure that can be divided into two major components: the dorsal thalamus and the ventral thalamus. The dorsal thalamus is made up of at least fifteen individual nuclei that project to the cerebral cortex, the outer layer of the cerebrum. The ventral thalamus, which consists mainly of the reticular nucleus, encapsulates the dorsal thalamus along the intercommissural line, providing clear landmarks to assist in defining the lateral and ventral borders of the dorsal thalamus. Being GABAergic, reticular cells modulate information by projecting into the dorsal thalamus to inhibit its relay cells (Pinault, 2004).
As most information flowing to the cortex passes through the thalamus first, the structure enjoys a strategic central position for brain processing (Guillery, 1995; Sherman & Guillery, 2006). The inter-thalamic adhesion is a flattened band of tissue that connects both parts of the thalamus at their medial surface. Mediodorsally, the wall of the third ventricle defines the thalamic border, while superiorly the lateral ventricle runs along the major portion of its rostro-caudal axis. On the lateral side, the dorsal thalamus is shielded by the reticular nucleus, a portion of the ventral thalamus that verges the internal capsule, which is white matter that separates the thalamus from the globus pallidus. Identifying the ventral border is complicated, especially so along its length, the rostro-caudal distance of the thalamus. Numerous structures are positioned inferiorly (e.g., the hypothalamus, the red nucleus and the substantia nigra), making it difficult to discern a clear border. Whereas the caudal end of the thalamus (the tip of the pulvinar) is mostly surrounded by ventricular space and thus easy to delineate, the rostral border of the thalamus is less well-defined, encouraging a caudal-rostral direction of segmentation (Power et al., 2015).
Connections and functions of thalamic nuclei
Generally viewed as a relay station of information from sensory organs and various subcortical areas to the cerebral cortex, the thalamus is a complex structure with a wide range of functions. Consisting of at least fifteen individual-nuclei, some of its many functions are the integration of (somato)sensory information, providing feedback to motor regions and regulating states of sleep and wakefulness. Interestingly, recent research has shown its role is not only limited to that of an information hub, but that it is also involved in the modulation of cortical functional networks (Hwang et al., 2017). Traditionally, the intra-thalamic nuclei are categorized according to their type of function, which results in three categories: (i) first order nuclei (specific), (ii) higher order nuclei (specific)
and (iii) non-specific nuclei. First order relay nuclei receive very well-defined, subcortical input and project this signal via ascending pathways to their interconnected, functionally distinct cortical targets and higher order thalamic relays (Guillery, 1995). These include the nuclei that transmit primary sensory information (the ventral posterolateral, ventral posteromedial, medial geniculate and lateral geniculate nuclei), the nuclei that provide a feedback pathway to specific motor regions (the ventral anterior and ventral lateral nuclei) and the anterior group. The second type, the higher order nuclei, are mutually interconnected to the association cortex. Higher order nuclei mostly receive input from the cerebral cortex and project back to the association areas of the cerebral cortex to modulate their activity (Guillery, 1995; Sherman & Guillery, 2006). These include the, the medial dorsal nucleus, the lateral dorsal and the lateral posterior nucleus and the pulvinar, the largest thalamic structure. The remaining nuclei are non-specific and include most of the intralaminar nuclei (the central lateral nucleus, the centromedian nucleus and the parafiscular nuclei), the midline nuclei and the reticular nuclei, of which the former two groups project diffusely through the cortex to regulate arousal and alertness. The reticular nucleus is the only nucleus that projects towards the dorsal thalamus itself, giving strong inhibitory input. Table 1 gives an overview of the thalamic nuclei, their connections and their functions. Note that most of the research on the connections and functions of thalamic nuclei is conducted on animals (e.g., rodents, non-human primates and cats). Even though, genetically, humans share high similarities with these animals (Mullins & Mullins, 2004; Rogers & Gibbs, 2014; Pontius et al., 2007), this lack of knowledge on the connections of the human thalamic nuclei further emphasizes the need for a detailed MRI-based map of the human subcortex.
Previous segmentation protocols for the thalamus
In the past decades, several protocols for thalamic segmentations on MRI have been established and validated. A number of these protocols are limited in their utility due to the lack of anatomical detail provided (Andreasen et al., 1990; Jernigan et al., 1991; Flaum et al., 1995; Gur et al., 1998; Lawrie et al., 1999; Dasari et al., 1999). However, there are a few that offer sufficient anatomical detail for other raters to utilize. Between these protocols, quality regarding the description of boundary definition varies. Portas and colleagues (1998), who compared thalamic volumes of schizophrenic patients with healthy controls, presenting a description of the thalamus in contiguous coronal 1.5 mm images, working in a rostral-caudal direction. In their protocol, they clearly report its anatomical landmarks, but information on exact
boundary definition is lacking. Spinks and colleagues (2002) worked in the same direction for manual segmentation of the thalamus, providing twelve coronal 1.5 mm sections. To make segmentation on a large-scale more viable, they used this protocol to train an artificial network to automatically define the thalamus and mediodorsal thalamus. To better discern landmarks, they used a multispectral image acquisition technique to acquire three different imaging types (T1, T2 and PD weighted images). Lastly, perhaps the most complete and detailed protocol was recently validated by Power and colleagues (2015). Working in a caudal-rostral direction, they provide a thorough description of the dorsal thalamus in 31 consecutive coronal 1.0 mm slices, acquired with T1-weighted 3T MRI, with precise boundary definition.
The aim of the current research is to improve on existing protocols, by establishing a protocol for manual segmentation of the thalamus on images obtained with 7T MRI. For this purpose, a new version of MP2RAGE (magnetization-prepared rapid gradient echo), called MP2RAGEME, is used. Different contrasts (e.g., T1w, T2*, R1, R2, QSM) will be visually inspected in order to find the best image contrast for the development of this protocol. Furthermore, where previous protocols included only coronal slices, this protocol will also add both axial and sagittal views to aid in defining indistinct boundaries, such as the lateral and inferior borders of the thalamus. For validation, the mean thalamic volumes (left, right and total) of the first and second round of segmentation will
be included for comparison. Inter-rater reliability will then be measured by calculating the correlation between the initial and second segmentation set of each side of the thalamus (left and right).
Methods
Participants
To establish the protocol, ten scans were randomly selected from a larger database acquired from the IMCN Research Unit of the University of Amsterdam, the Netherlands, as part of an ongoing project to map the human subcortex. These scans were analysed and used to train the rater to recognize anatomical landmarks and manually segment the thalamus. From these scans, images in various views (coronal, sagittal and axial) were included to the protocol. Once the protocol was in a finalized state, the thalami of six scans were segmented for validation (mean age 33, ranged between 20 -66; 2 males, 4 females). Written consent was given by all participants. The study was approved by the Local Ethics Committee at the University of Amsterdam.
Image acquisition and volumetric analysis
MRIs were acquired using an ultra-high field 7.0 Tesla Philips MR scanner, with a 32-channel head coil (Philips 7T
Table 1. Classification, connections and functions of the thalamic nuclei. Nuclei Primary afferent
connection(s) Primary efferent connection(s) Function(s) References
Anterior
group Hippocampal formation (subiculum and presubiculum) via the mammillary bodies
Posterior cingulate gyrus; retrosplenial area; subiculum and presubiculum
Limbic relay (related to modulation of alertness, emotional learning and memory functions)
Dere et al., 2008; Child & Benarroch, 2013**** Lateral
geniculate nucleus (LGN)
Optic tract (retinotopic input, ganglion cells); primary visual cortex (corticothalamic feedback)
Primary visual cortex
(calcarine fissure) Visual relay Goodale & Milner, 2004**; Goodale & Milner, 2006**; Cudeiro & Sillito, 2006; O’Connor et al., 2002****; Schneider et al., 2004**** Medial geniculate nucleus (MGN)
(Brachium of) inferior colliculi Primary auditory cortex
(superior temporal gyrus) Auditory relay Winer, 1992; Puelles et al., 2012* Ventral
anterior nucleus (VA)
Medial globus pallidus; substantia
nigra (pars reticula) Prefrontal cortex; cingulate; premotor cortex; supplementary motor area
Modulation of motor function
through feedback Jones, 2007; Fitzgerald, 2012; Swenson, 2012; Xiao & Barbas, 2004**
Ventral lateral nucleus (VL)
Denate nucleus of cerebellum; vestibular nuclei; spinothalamic tract; medial globus pallidus
Primary motor cortex; premotor cortex;
supplementary motor area
Modulation of motor function through feedback; coordination and initiation of movements Jones, 2007; Orrison, 2008 Ventral posterolater al nucleus (VPL)
Sensory tracts (spinothalamic
tracts; medial memniscus) Primary somatosensory cortex(medial two-thirds of the postcentral gyrus); secondary somatosensory cortex
Tactile sensations (pressure; pain; temperature) and proprioceptive sensations (trunk and limbs)
Jones, 2007; Gauriau & Bernard, 2004*; Tracey, 2004*; Waite, 2004*; Francis et al., 2008* Ventral posteromedi al nucleus (VPM) Sensory tracts
(trigeminothalamic tracts) from the head region; solitary tracts; substantia nigra
Somatosensory cortex (inferolateral portion of the postcentral gyrus); cortical gustatory area; frontal areas
Tactile sensations (pressure; pain; temperature) and proprioceptive sensations (head); gustation
Jones, 2007; Gauriau & Bernard, 2004*; Tracey, 2004*; Waite, 2004*; Arthurs & Reilly, 2013* Lateral posterior (LP)
Superior colliculus; parietal
cortex; pretectum Parietal association cortex Integration of (somato)sensory information Jones, 2007 Lateral
Nuclei Primary afferent
connection(s) Primary efferent connection(s) Function(s) References
Medial dorsal nucleus (MD)
Amygdala; olfactory cortex; septal area; hypothalamus; limbic system; superior colliculus
Frontal areas (prefrontal cortex, orbitofrontal and lateral frontal); olfactory structures; limbic system
Limbic relay to frontal areas;
olfactory relay Jones, 2007; Goldman-Rakic & Porrino, 1985**; Armstrong & Paxinos, 1990****; Hirai & Jones, 1989****; Li et al., 2004* Pulvinar Superior colliculus; pretectum;
association cortex; auditory cortex; visual cortex
Parietotemporal area (secondary visual areas & association areas); prestriate cortex
Facilitation of intra- and cross-modal cortical information processing; integration of sensory information (attentional engagement, visual filtering & orientation to feature changes)
Jones, 2007; Romanski et al., 1997**; Berman & Wurtz, 2011**; Petersen et al., 1987**; Chalupa, 1991**; Arend et al., 2008****; Rafal & Posner, 1987****; LaBerge & Buchsbaum, 1990****; Michael & Buron, 2005**** Central
lateral nucleus (CL)
Reticular formation; substantia nigra; cerebellar nuclei; superior colliculus; spinothalamic tract (nociceptive input)
Wide areas of cerebral cortex (motor cortex, somatosensory cortex, parietal cortex & frontal cortex); corpus striatum (putamen)
Regulation of excitatory levels of cerebral cortex; arousal; modulation of states of sleep and wakefulness; eye movement; nociception
Jones, 2007; Van der Werf et al., 2002*
Central median nucleus (CM)
Reticular formation; substantia nigra; cerebellar nuclei; superior colliculus
Corpus striatum; parietal
cortex; frontal cortex Regulation of excitatory levels of cerebral cortex; arousal; modulation of states of sleep and wakefulness; eye movement; nociception
Van der Werf et al., 2002*; Royce et al., 1991***; Yamanaka et al., 2017**; Powell & Cowan, 1967** Parafiscular
nucleus Reticular formation; periaqueductal grey Frontal cortex Regulation of excitatory levels of cerebral cortex; arousal; modulation of states of sleep and wakefulness; eye movement; nociception
Van der Werf et al., 2002*; Yamanaka et al., 2017**; Royce et al., 1991***; Reuniens
nucleus Fornix Hippocampal formation; amygdala; cingulate; hypothalamus
Viscero-limbic functions;
attentiveness Vertes & Hoover, 2006*; McKenna & Vertes, 2004*; Wouterlood et al., 1990*; Vertes et al., 2007* Paratenial
nucleus A large number of limbic and limbic-associated structures; brainstem
Medial frontal cortex Limbic functions Jones, 2007; Vertes, 2008*
Reticular
nucleus Reticular formation; thalamic nuclei; cerebral cortex Dorsal thalamic nuclei (Inhibitory) relay of signals through thalamus; setting thalamocortical circuits in motion
Jones, 1975*,**,***; Crabtee
& Isaac, 2002*; Guillery & Harting, 2003; Pinault, 2004*,**,***
Note. * = RODENTS, ** = NON-HUMAN PRIMATES, *** = CATS, **** = HUMANS
Table 2. Mean thalamic volumes.
Left thalamus (N = 6) Right thalamus (N = 6) Total thalamus (N = 6) Mean (mm³ x 10³) S.D. (mm³ x 102) Mean (mm³ x 10³) S.D. (mm³ x 102) Mean (mm³ x 103) S.D. (mm³ x 102) First set 6.302 2.396 6.5 3.003 12.8021 5.379 Second set 6.138 3.270 6.3462 2.651 12.4842 5.639 S.D. = standard deviation.
Achieva, Philips Medical Systems, Best, The Netherlands). For optimized contrast, a novel version of the MP(2)RAGE (Marques et al., 2014; Mugler & Brookeman, 1990) sequence was used. High resolution images were acquired using a MP2RAGEME pulse sequence. MP(2)RAGE has become one of the most widely used sequences to acquire T1-weighted (T1w) images of the human brain with fine grey matter (GM)/white matter (WM) contrast (Ashburner & Friston, 2000). Several contrasts (T1w, T2*map, R1, R2, QSM) were visually inspected and compared to determine which is the most suitable for visualization and segmentation of the thalamus.
Images were acquired using a MP2RAGEME sequence with the following parameters: repetition time (TR) 6298 ms; echo times (TE1) 3 ms and (TE2) 3 ms; inversion times (TI1) 670 ms and (TI2) 2325.5 ms; flip angles (FA1) 4° and (FA2) 4°; bandwidth 476 Hz; SENSE factor 2 (PA); the acquisition matrix was 292 x 290 with a field of view (FOV) of 205 x 205 x 164 mm;acquired voxel size 0.7 x 0.7 x 0.7 mm³; the reconstructed voxel size 0.64 x 0.64 x 0.7 mm3.
On the second inversion multiple echos were added, starting with a TE of 3, increasing delta-echo time of 8.5 (3, 11.5, 19, 27.5). Acquisition time was approximately 20 minutes. The MP2RAGE images were anonymized. Unilateral manual segmentation of the thalami of the six subjects and volume calculation were performed by an intern of the Integrative Model-based Cognitive Neuroscience (IMCN) Research Unit. A standard sectional anatomical atlas (Mai, Paxinos & Voss, 2007) was used for reference and assistance during segmentation (see figure 1 for page example). For counterbalancing purposes, the starting hemisphere was switched between each subject. Segmentation was performed using FslView 4.0.1 software.
Volumetric and statistical analysis
Mean volumes were calculated of the left thalamus, the right thalamus and the total thalamus. Intra-rater reliability is assessed by calculating the Pearson correlation between the volumes measured in the first and second set, for the left and the right thalamus. The calculation was done in SPSS 20.0 (International Business Machines Corporation, NY, USA).
Results
After visual inspection of various image contrasts, the T1-weighted contrast was chosen as the most suitable for identification and delineation of the thalamus. This is in line with previous studies reporting a protocol for manual
Table 3. Intra-rater reliability.
Correlation coefficient
Left thalamus 0.91*
Right thalamus 0.94**
Pearson correlation coefficients for twelve measurements of the thalamus (six per each side). *p < 0.05, **p < 0.01.
segmentation of the thalamus (Tourdias, 2014; Power et al., 2015). The segmentation protocol created with these images is added to this report as the appendix. Once the protocol was finalized, it was used to segment the thalami of six scans. Mean thalamic volumes (left thalamus, right thalamus and total thalamic volume) were calculated and are given in Table 2. Intra-rater reliability was determined for six thalami per each side. The correlation was 0.91 for the left thalamus and 0.94 for the right thalamus (see Table 3).
Discussion
The goal of this study was to provide a protocol for manual segmentation of the
thalamus, as part of a greater project by the IMCN Research Unit to create a detailed atlas of the human subcortex.The segmentation protocol presented in this report improves on previously created protocols for the thalamus on a number of levels. First, the protocol has been established with the use of state-of-the-art neuroimaging techniques at a high resolution with slices of less than 1 mm thick. The novel MP2RAGE(ME) sequence provides an enhanced contrast, to optimize border distinction. Second, working in a caudal to rostral direction, it contains 43 images of coronal slices, significantly more comparing to previously developed protocols and matching up to the protocol made by Power and colleagues (2015). Next to these images, both axial and sagittal views are provided as well. These are added to assist in boundary definition, especially where boundaries are not immediately distinguishable (such as the lateral border, the ventral border and, in some sections, the dorsal border). Furthermore, the protocol refers to a detailed neurosurgical atlas (Mai, Paxinos & Voss, 2007) for verification of arbitrary landmarks. Lastly, whereas not validated by means of segmentation due to time constraints, the protocol includes the description and distinction of a subset of thalamic structures, such as the pulvinar nucleus and the mediodorsal thalamus.
Several studies reporting a segmentation protocol for the thalamus have included mean thalamic volumes. This allows for a comparison of the volumes found in the current study with the ones reported in these studies. Lawrie and colleagues (1999) reported mean values of 6.60 cm3 (S.D. ± 0.80) for the left thalamus and 6.40
cm3 (S.D. ± 0.60) for the right thalamus in
healthy control subjects (mean age 21.1 years, S.D. ± 2.3). Ettinger and others (2001) followed the protocol made by Portas and colleagues (1998) and reported an unadjusted mean total volume of the thalamus of 15.10 cm3 in a healthy
control group (mean age 25.4 years, S.D. ± 5.8). van
Figure 1. Page example from the MAI atlas (Mai, Paxinos & Voss, 2007), showing a coronal section of the thalamus and surrounding subcortical structures.
der Werf et al. (2001) measured mean thalamic volumes across time, reporting a significant decrease in total volume as age increased in healthy individuals. They found unadjusted mean volumes of 15.29 cm3 for subjects aged between
22-45 years (n = 18), 13.41 cm3 for the group
aged between 46-60 years (n = 14) and 12.22 cm3 for subjects aged between 61-82 years (n =
24). Important to note is that the thalamic age-related volume decrease found in these subjects were visible before the onset of total brain volume loss was apparent. Spinks and others (2002) measured mean values of 7.00 cm3 (S.D.
± 1.12) for the left thalamus and 6.55 cm3 (S.D.
± 0.54) for the right thalamus in healthy males (mean age 25.63 years, S.D. ± 6.55). Lastly, Power et al. (2015) reported mean values for the left thalamus of 5.59 cm3 (S.D. ± 4.11) by the
first rater, and 5.51 cm3 (S.D. ± 4.54) and 5.66
cm3 (S.D. ± 4.11) by the second rater. For the
right thalamus, their first rater reported a mean value of 5.67 cm3 (S.D. ± 3.4), and the second
rater reported mean values of 5.80 cm3 (S.D. ±
4.2) and 5.66 cm3 (S.D. ± 3.4). Mean total
thalamic volumes were 11.26 cm3 (S.D. ± 7.44)
reported by the first rater and 11.31 cm3 (S.D. ±
8.68) and 11.32 cm3 (S.D. ± 7.42) by the second
rater (mean age 56, ranged between 35-72). The mean volumes from the current study (see Table 2) compare well with those reported in most previous studies. The only study reporting substantially smaller mean thalamic volumes is the one by Power et al. (2015), implying that a more liberal method of segmentation may have been employed using the current protocol. It is
worth mentioning, however, that the subject group in the study by Power et al. is considerably older (mean age 56, ranged between 35 – 72), which could partially explain the smaller volumes reported in their study. This is supported by the results found by van der Werf et al. (2001), indicating a decrease of thalamic volume with an increase in age.
This protocol has a number of limitations. While intra-rater correlations of 0.91 and 0.94 indicate a high reliability, these coefficients are based on only six scans. Validation should be improved by increasing the number of segmented scans for the validation of this protocol. In addition, an inter-rater coefficient should be included to confirm reliability among multiple raters. A strong and commonly used statistical validation metric for inter-rater variability is the Dice similarity coefficient (Dice, 1945). This statistic can be used to evaluate the reproducibility and spatial overlap of manual segmentations, also accounting for the volume of the structure. Furthermore, what is important to note is that the reporting of structure volume does not suffice for the differentiation between various thalami. Next to structural MRI, other modalities such as shape analysis (e.g., Spherical Harmonics) and diffusion-weighted tensor imaging (DTI) should be used. These tools will give additional information on (statistically significant) volume differences per region and cortical connectivity of different thalamic nuclei respectively. Another option would be to use the Hausdorff distance, a measure for the
dissimilarity of two shapes, taking volume into account as well (Henrikson, 1999).
As previously stated by Power et al. (2015) and confirmed during the development of this protocol, the lateral boundary of the thalamus remains blurry, even so at 7T MRI. The reticular thalamic nucleus, part of the ventral thalamus, is a thin structure encapsulating a major portion of the dorsal thalamus along this border. Because the reticular nucleus differs functionally from the rest of the thalamus (it does not project to the cerebral cortex), we attempted to exclude this structure. However, the unclear border of the dorsal thalamus with the reticular thalamic nucleus makes exact delineation challenging. In some slices, the reticular nucleus may not have been completely excluded. It is because of arbitrary boundaries such as these that manual segmentation of the thalamus in MRI should always be executed by an experienced rater. Secondly, the lateral geniculate nucleus (LGN) is separated from the rest of the dorsal thalamus in several coronal slices, which is problematic for shape analysis. For improved versions of this protocol, the research unit has decided to cut off the discontinuous LGN in these slices. Lastly, it should be noted that there was a learning curve during the development of this protocol. The protocol contains images of segmentations done in the early stage of this project. On these scans, in rostral sections especially, the segmentation in the protocol images may have been executed too liberally at the lateral and inferior borders. More specifically, part of the hypothalamus may have been included in the mask of the images. Consulting a detailed neurosurgical atlas (e.g., the Mai atlas) while delineating these boundaries is therefore highly recommended.
The aim of this protocol specifically is to aid in the investigation of how the shape and volume of the thalamus relates to neural connections that underlie human cognitive functioning, the integration of sensory information, emotional processing, motor control and behaviour. While this protocol serves as a detailed step-by-step guidance for boundary definition of the thalamus and provides a description of some of the thalamic nuclei, it is still in a preliminary stage. With the fine contrast that is acquired with the MP2RAGEME pulse sequence at high resolution obtained with 7T, clear visualization of a number of these substructures is possible. In the future, the protocol could be improved by including detailed steps for segmentation of various thalamic nuclei, in a similar fashion as has been done for the whole thalamus in this protocol. Acquiring detailed neuroanatomical maps of these subcortical structures and registering these maps into standard space would greatly facilitate future functional MRI research. Using a probabilistic atlas and thus being able to account for certain demographics such as age and sex may assist in determining how these factors relate to human
thalamic functioning. Using the newest techniques and high resolution images, developing detailed protocols such as the one presented in this report will lead us towards a better understanding of the mechanisms and functions of the human subcortex.
References
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, 50.
Andreasen, N. C., Ehrhardt, J. C., Swayze, V. W., Alliger, R. J., Yuh, W. T. C., Cohen, G. & Ziebell, S. (1990). Magnetic resonance imaging of the brain in schizophrenia. Archives of General Psychiatry, 47, 35 – 44.
Andreasen, N. C., Arndt, S., Swayze, V., Cizadlo, T., Flaum, M., O’Leary, D. S., Ehrhardt, J. C. & Yuh, W. T. C. (1994). Thalamic abnormalities in schizophrenia visualized through magnetic resonance image averaging. Science, 266,
294 – 298.
Annese, J. (2012). The importance of combining MRI and large-scale digital histology in neuroimaging studies of brain connectivity and disease. Frontiers in Neuroinformatics, 6(13), 1 – 6.
Arend, I., Rafal, R. & Ward, R. (2008). Spatial and temporal deficits are regionally dissociable in patients with pulvinar lesions. Brain, 131, 2140 – 2152. doi:10.1093/brain/awn135
Armstrong, E. & Paxinos, G. (1990). The limbic thalamus: anterior and mediodorsal nuclei. In G. Paxinos (Ed.), The human nervous system (3rd ed.), (pp. 469 – 478). San Diego, SD:
Academic Press.
Arthurs, J. & Reilly, S. (2013). Role of the gustatory thalamus in taste learning.
Behavioural Brain Research, 250, 9 – 17. Ashburner, J. & Friston, K. J. (2000). Voxel-based
morphometry—the methods. NeuroImage,
11, 805 – 821.
Berman, R. & Wurtz, R. (2011). Signals conveyed in the pulvinar pathway from superior colliculus to cortical area MT. The Journal of
Neuroscience, 31, 373 – 384. doi: 10.1523/jneurosci.4738-10.2011.
Bridge, H. & Clare, S. (2006). High-resolution MRI: in vivo histology? Philosophical
Transactions of the Royal Society B: Biological Sciences, 361(1465), 137 – 146.
Buchsbaum, M. S., Someya, T., Teng, C. Y., Abel, L., Chin, S., Najafi, A., Haier, R. J., Wu, J. & Bunney, W. E. (1996). PET and MRI of the thalamus in never medicated patients with schizophrenia. American Journal of Psychiatry,
153, 191 – 199.
Chalupa, L. M. (1991). Visual function of the pulvinar. In A. G. Leventhal (Ed.), Vision and
Visual Function, Vol: 4. CRC Press, Boca Raton,
FL., pp. 140 – 159.
Child, N. D. & Benarroch, E. E. (2013). Anterior nucleus of thethalamus: functional organization and clinical implications. Neurology, 81, 1869 –
1976. doi:
10.1212/01.wnl.0000436078.95856.56. Crabtree, J. W. & Isaac, J. T. (2002). New intrathalamic pathways allowing modality-related and cross-modality switching in the dorsal thalamus. Journal of Neuroscience, 22, 8754 – 8761.
Cudeiro, J. & Sillito, A. M. (2006). Looking back: corticothalamic feedback and early visual processing. Trends in Neurosciences, 29, 298 – 306. doi: 10.1016/j.tins.2006.05.002. Dasari, M., Friedman, L., Jesberger, J., Stuve, T. A., Findling, R. L., Swales, T. P. & Schulz, S. C. (1999). A magnetic resonance imaging study of thalamic area in adolescent patients with either schizophrenia or bipolar disorder as compared to healthy controls. Psychiatry Research: Neuroimaging, 91, 155 – 162.
Dere, E., Easton, A., Nadel, L. & Huston, J. (2008).
Handbook of Episodic Memory. Retrieved from
http://books.google.com
Dice, L. R. (1945). Measures of the amount of ecologic association between species.
Ecology, 26, 297 – 302.
Duyn, J. H. (2012). The future of ultra-high field MRI and fMRI
for study of the human brain.
Neuroimagine, 62, 911 – 922. doi: 10.1016/j.neuroimage.2012.01.024
Ettinger, U., Chitnis, X. A., Kumari, V., Fannon, D. G., Sumich, A. L., O’Ceallaigh, S., Doku, V. C. & Sharma, T. (2001). Magnetic resonance imaging of the thalamus in first- episode psychosis. The American Journal of Psychiatry,
158, 116 – 118.
Evans, A. C., Janke, A. L., Collins, D. L. & Baillet, S. (2012).
Brain templates and atlases. Neuroimage,
62, 911 – 922. doi:
10.1016/j.neuroimage.2012.01.024
Fitzgerald, M. J. T. (2012). Clinical Neuroanatomy
and Neuroscience. Retrieved from
http://books.google.com
Flaum, M., Swayze, V. W., O’Leary, D. S., Yuh, W. T. C., Ehrhardt, J. C., Arndt, S. V. & Andreasen, N. C. (1995). Effects of diagnosis, laterality, and
gender on brain morphology in
schizophrenia. American Journal of Psychiatry,
152, 704 – 714.
Francis, J. T., Xu, S. & Chapin, J. K. (2008). Proprioceptive and cutaneous representations in the rat ventral posterolateral thalamus.
Journal of Neurophysiology, 99, 2291 – 2304.
Gauriau, C. & Bernard, J. F. (2004) A comparative reappraisal of projections from the superficial laminae of the dorsal horn in the rat: the forebrain. Journal of Comparative Neurology,
468, 24 – 56.
Goldman-Rakic, P. S. & Porrino, L. J. (1985). The primate mediodorsal (MD) nucleus and its
projection to the frontal lobe. Journal of
Comparative Neurology, 242, 535 – 560. Goodale, M. & Milner, D. (2004). Sight unseen: An
Exploration of Conscious and Unconscious Vision. Retrieved from
http://books.google.com
Goodale, M. & Milner, D. (2006). The Visual Brain
in Action. New York, NY: Oxford University Press.
Guillery, R. W. (1995). Anatomical evidence concerning the role of the thalamus in corticocortical communication: A brief review. Journal of Anatomy, 187, 583 - 592.
Guillery, R. W. & Harting, J. K. (2003). Structure and connections of the thalamic reticular nucleus: advancing views over half a century. Journal of Comparative Neurology, 463, 360 – 371.
Gur, R. E., Maany, V., Mozley, P. D., Swanson, C., Bilker, W. & Gur, R. C. (1998). Subcortical MRI volumes in neuroleptic-naıve and treated patients with schizophrenia. American Journal of
Psychiatry, 155, 1711 – 1717.
Henrikson, J. (1999). Completeness and total boundedness of the Hausdorff metric. MIT
Undergraduate Journal of Mathematics, 1, 69
-80.
Hirai, T. & Jones, E. G. (1989). A new parcellation of the humanthalamus on the basis of histochemical staining. Brain Research
Reviews, 14, 1 – 34.
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, e0120572. Hwang, K., Bertolero, M., Liu, W. & D’Esposito, M. (2017). The
human thalamus is an integrative hub for functional brain networks. The Journal of
Neuroscience, 1 - 20. doi: 10.1523/jneurosci.0067- 17.2017
Jernigan, T. L., Archibald, S. L., Berhow, M. T., Sowell, E. R., Foster, D. S. & Hesselink, J. R. (1991). Cerebral structure on MRI, Part I: localization of age-related changes. Biological
Psychiatry, 29, 55 – 67.
Jones, E. G. (1975). Some aspects of the organization of the thalamic reticular complex.
Journal of Comparative Neurology, 162, 285
– 308. doi: 10.1002/cne.901620302.
Jones, E. G. (2007). The Thalamus (2nd ed.).
Cambridge, UK: Cambridge University Press. LaBerge, D. & Buchsbaum, M. S. (1990). Positron emission tomographic measurements of pulvinar activity during an attention task.
Journal of Neuroscience, 10, 613 – 619.
Lawrie, S. M., Whalley, H., Kestelman, J. N., Abukmeil, S. S., Byrne, M., Hodges, A., Rimmington, J. E., Best, J. J. K., Owens, D. G. C. & Johnstone, E. C. (1999). Magnetic resonance imaging of brain in people at high risk of developing schizophrenia. Lancet,
353, 30 – 33.
Li, X. B., Inoue, T., Nakagawa, S. & Koyama, T. (2004). Effect of mediodorsal thalamic nucleus lesion on contextualfear conditioning in
rats. Brain Research, 1008, 261 – 272. doi: 10.1016/j.brainres.2004.02.038.
Mai, J. K., Paxinos, G. & Voss, T. (2008). Atlas of
the Human Brain (3rd ed.). San Diego, SD:
Academic Press.
Marques, J. P., Kober, T., Krueger, G., van der Zwaag, W., van de Moortele, P. F. F. & Gruetter, R. (2010). MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. Neuroimage, 49, 1271 – 1281.
Maubon, A., Ferru, J., Berger, V., Soulage, M. C., DeGraef, M.,
Aubas, P., Coupeau, P., Dumont, E. & Rouanet, J. P. (1999). Effect of field strength on MR images: comparison of the same subject at 0.5, 1.0, and 1.5 T. Radiographics, 19, 1057 – 1067.
McKenna, J. T. & Vertes, R. P. (2004). Afferent projections to nucleus reuniens of the thalamus.
Journal of Comparative Neurology, 480, 115 –
142. doi:10.1002/cne.20342.
Michael, G. A. & Buron, V. (2005). The human pulvinar and stimulus-driven attentional control.
Behavoural Neuroscience, 119, 1353 – 1367.
Mugler, J. P. I. & Brookeman, J. R. (1990). Three-dimensional magnetization- prepared rapid gradient-echo imaging (3D MP-RAGE).
Magnetic Resonance in Medicine, 15, 152
– 157.
Mullins, L. J. & Mullins, J. J. (2004). Insights from the rat genome sequence. Genome
Biology, 5, 221.
O'Connor, D. H., Fukui, M. M., Pinsk, M. A. & Kastner, S. (2002). Attention modulates responses in the human lateral geniculate nucleus. Nature Neuroscience, 5, 1203 – 1209. Orrison, W. (2008). Atlas of Brain Function. New York, NY: Thieme Medical Publishers.
Osechinskiy, S. & Kruggel, F. (2009). Quantitative comparison of high-resolution MRI and myelin-stained histology of the human cerebral cortex. Annual International Conference
Proceedings of the IEEE Engneering in Medicine and Biology Society, 85 – 89.
Petersen, S., Robinson, D. & Morris, J. (1987). Contributions of the pulvinar to visual spatial attention. Neuropsychologia, 25, 97 – 105.
doi: 10.1016/0028- 3932(87)90046-7.
Pinault, D. (2004). The thalamic reticular nucleus: structure, function and concept. Brain
Research Reviews, 46, 1 – 31.
Pontius, J. U., Mullikin, J. C., Smith, D. R., Lindblad-Toh, K., Gnerre, S., Clamp, M., Chang, J., Stephens, R., Neelam, B., Volfovsky, N., Schäffer, A. A., Agarwala, R., Narfström, K., Murphy, W. J., Giger, U., Roca, A. L., Antunes, A., Menotti-Raymond, M., Yuhki, N., Pecon-Slattery, J., Johnson, W. E., Bourque, G., Tesler, G. & O’Brien, S. J. (2007). Initial sequence and comparative analysis of the cat genome. Genome Research, 17, 1675 –
1689. doi: 10.1101/gr.6380007.
Portas, C. M., Goldstein, J. M., Shenton, M. E., Hokama, H. H., Wible, C. G., Fischer, I., Kikinis, R., Donnino, R., Jolesz, F. A. &
MCarley, R. W. (1998). Volumetric evaluation of the thalamus in schizophrenic male patients using magnetic resonance imaging. Biological Psychiatry, 43, 649 – 659.
Powell, T. P. S. & Cowan, W. M. (1967). The interpretation of the degenerative changes in the intralaminar nuclei of the thalamus.
Journal of Neurology, Neurosurgery & Psychiatry, 30, 140 – 153. doi: 10.1136/jnnp.30.2.140.
Power, B. D., Wilkes, F. A., Hunter-Dickson, M., van Westen, D., Santillo, A. F., Walterfang, M., Nilsson, C., Velakoulis, D. & Looi, J. C. L. (2015). Validation of a protocol for manual segmentation of the thalamus on magnetic resonance imaging scans.
Psychiatry Research, 232, 98 – 105.
Puelles, L., Martinez-de-la-Torre, M., Ferran, J. & Watson, C. (2012). Diencephalon. In C. Watson, G. Paxinos & L. Puelles (Eds.), The
Mouse Nervous System (1st ed.) (pp. 313 – 336). San Diego, SD: Academic Press.
Rafal, R. D. & Posner, M. I. (1987). Deficits in human visual spatial attention following thalamic lesions. Proceedings of the National Academy of Sciences, 84, 7349 – 7353.
Rogers, J., & Gibbs, R. A. (2014). Comparative primate genomics: emerging patterns of genome content and dynamics. Nature Reviews
Genetics, 15, 347 – 359. http://doi.org/10.1038/nrg3707
Romanski, L. M., Giguere, M., Bates, J. F. & Goldman-Rakic, P. S. (1997). Topographic organization of medial pulvinar connections with the prefrontal cortex in the rhesus
monkey. Journal of Comparative
Neurology, 379, 313 - 332.
Royce, G. J., Bromley S., Gracco, C. (1991). Subcortical projections to the centromedian and parafascicular thalamic nuclei in the cat.
Journal of Comparative Neurology, 306, 129
– 155.
Rutt B. K., Lee, D. H. (1996). The impact of field strength on
image quality in MRI. Journal of Magnetic
Resonance Imaging, 6, 57 - 62.
Schafer, A., Forstmann, B. U., Neumann, J., Wharton, S., Mietke, A., Bowtell, R. & Turner, R. (2012). Direct visualization of the subthalamic nucleus and its iron distribution using high-resolution susceptibility mapping.
Human Brain Mapping, 33, 2831 – 2842. doi:
10.1002/hbm.21404.
Schneider, K. A., Richter, M. C. & Kastner, S. (2004). Retinotopic Organization and Functional Subdivisions of the Human Lateral Geniculate Nucleus: A High- Resolution Functional Magnetic Resonance Imaging Study. The
Journal of Neuroscience, 24, 8975 – 8985.
Sherman, S. M. & Guillery, R. W. (2006). Exploring
the Thalamus and its Role in Cortical Function.
Retrieved from http://brainmaster.com Spinks, R., Magnotta, V. A., Andreasen, N. C., Albright, K. C., Ziebell, S., Nopoulos, P. & Cassell, M. (2002). Manual and automated measurement of the whole thalamus and
mediodorsal nucleus using magnetic resonance imaging. Neuroimage, 17, 631–642.
Swenson, R. S. (2006). Review of clinical and
functional neuroscience. Retrieved from
https://www.dartmouth.edu/~rswenson/NeuroSci/ Talairach, J. & Tournoux, P. (1998). Co-planar
Stereotactic Atlas of the Human Brain: 3-dimensional Proportional System: An Approach to Cerebral Imaging. Retrieved from
http://afni.nimh.nih.gov/afni/.
Tourdias, T., Saranathan, M., Levesque, I. R., Su, J. & Rutt, B. K. (2014). Visualization of intra-thalamic nuclei with optimized white-matter-nulled MPRAGE at 7 T. NeuroImage, 84, 534
– 545.
Tracey, D. (2004). Somatosensory system. In G. Paxinos (Ed.), The Rat Nervous System (3rd ed.), (pp. 797 – 815). San Diego, SD: Academic Press.
Van der Werf, Y. D., Tisserand, D. J., Visser, P. J., Hofman, P. A. M., Vuurman, E., Uylings, H. B. M. & Jolles, J. (2001). Thalamic volume predicts performance on tests of cognitive speed and decreases in healthy aging. A magnetic resonance imaging-based volumetric
analysis. Cognitive Brain Research, 11, 377 – 385.
Van der Werf, Y.D., Witter, M. P. & Groenewegen, H. J. (2002). The intralaminar and midline nuclei of the thalamus. Anatomical and functional evidence for participation in processes of arousal and awareness. Brain Research Reviews, 39, 107 – 140.
Vertes, R. P. (2006). Interactions among the medial prefrontal cortex, hippocampus and midline thalamus in emotional and cognitive processing in the rat. Neuroscience, 142, 1 – 20. doi: 10.1016/j.neuroscience.2006.06.027. Vertes, R. P. & Hoover, W. B. (2008). Projections of the paraventricular and paratenial nuclei of the dorsal midline thalamus in the rat. Journal
of Comparative Neurology, 508, 212 – 237.
doi: 10.1002/cne.21679.
Vertes, R. P., Hoover, W. B., Szigeti-Buck, K. & Leranth, C. (2007). Nucleus reuniens of the midline thalamus: Link between the medial prefrontal cortex and the hippocampus.
Brain Research Bulletin, 71, 601 – 609.
E. Visser, M. Keuken, G. Douaud, V. Gaura, A. C. Bachoud-Levi, P. Remy, B. Forstmann & M. Jenkinson (2016). Automatic segmentation of the striatum and globus pallidus using MIST: multimodal image segmentation tool.
NeuroImage, 125, 479 – 497.
Waite, P. M. E. (2004). Trigeminal sensory system. In G. Paxinos, (Ed.), The Rat Nervous System (3rd ed.), (pp. 817 – 851). San Diego, SD: Academic Press.
Winer, J. A. (1992). The functional architecture of the medial geniculate body and primary auditory cortex. In D. B. Webster, A. N. Popper & R. R. Fay (Eds.), The Mammalian Auditory Pathway: Neuroanatomy (1st ed.), (pp. 222 – 409). New York, NY: Springer-Verlag.
Wouterlood, F. G., Saldana, E. & Witter, M. P. (1990). Projectio from the nucleus reuniens thalami to the hippocampal region: Light and electron microscopic tracing study in the rat with the anterograde tracer Phaseolus vulgaris- leucoagglutinin. Journal of Comparative Neurology, 296, 179 – 203. doi:
10.1002/cne.902960202.
Xiao, D. & Barbas, H. (2004). Circuits through prefrontal cortex, basal ganglia, and ventral anterior nucleus map pathways beyond motor control. Thalamus & Related
Systems, 2, 325 – 343.
Yamanaka, K., Hori, Y., Minamimoto, T., Yamada, H., Matsumoto, N., Enomoto, K., Aosaki, T., Graybiel, A. M. & Kimura, M. (2017). Roles of centromedian parafascicular nuclei of thalamus and cholinergic interneurons in the dorsal striatum in associative learning of environmental events. Journal of Neural
Transmission, 1 – 13.
APPENDIX
Segmentation Protocol of the Thalamus using MP2RAGEME at 7T by Simon Poortman
1. Introduction
1.1 MR sequence and contrast
The type of sequence is a Magnetization Prepared 2 Rapid Gradient Echo (MP2RAGE) pulse sequence, with an extension in which multiple-echos were added to the second inversion (MP2RAGEME). The parameters set as follows: repetition time (TR) 6298 ms; echo times (TE1) 3 ms and (TE2) 3 ms; inversion times (TI1) 670 ms and (TI2) 2325.5 ms; flip angles (FA1) 4° and (FA2) 4°.
2. Step-by-step delineation of the thalamus in caudal to rostral progression
Figure 1. Axial views (top) and sagittal views (bottom) depicting the border between the
very caudal part of the thalamus and the superior cistern.
Considering the relatively distinct thalamic border in the most caudal sections and the less distinguishable thalamic boundary on the rostral end, this protocol works in a caudal to rostral (towards the nose) orientation.
To pinpoint the most caudal portion of the thalamus (the tail end of the pulvinar), it is advised to do so with the assistance of either an axial or a sagittal slice. This provides a clear contrast of the border of the thalamus and the superior cistern, which is ventricular space (see figure 1).
Figure 2. Two identical coronal sections of the brain. In the left version, the caudal portion
of the thalamus (pulvinar) is marked red, with relevant surrounding landmarks annotated. Starting caudally, the first portion of the thalamus which can be visualized is the pulvinar. In a coronal section, this part can be recognized by its oval shape, lying between the crus fornix (white matter tract) diagonally running along its superolateral side and the superior cistern (ventricular space) situated medio-inferiorly. The pineal gland is positioned centrally, superior to the colliculi. Keep in mind that in the first number of sections, the thalamic mass increases in size quickly (see figure 2).
Figure 4.
Moving on into the rostral direction, the pulvinar gradually becomes more distinguishable. The crus fornix will be slightly less defined and thus delineating the roof of the thalamus may become more difficult. Consulting a sagittal view of the thalamus to define the superior border may prove helpful. As can be seen in both figure 3 and 4, the superior cistern is still clearly visible and the midbrain has not merged with the medio-inferior part of the thalamus yet. Note the tail of the caudate nucleus, superolateral to the crus fornix, at the lateral edge of the lateral ventricle (see figure 4).
Figure 6.
Figure 7.
Boundary definition is still relatively easy due to its oval shape and clear borders with ventricular space medially and inferiorly.
Figure 8b.
The central aqueduct, which is still evident here, can be used a landmark for orientation (see figure 8a). The lateral border of the thalamus is becoming more defined by the white matter that will develop into the internal capsule. This may cause the exact boundary to become somewhat arbitrary at times. An axial view can help for the definition of the lateral border along the entire length of the thalamus (see figure 8b for an example). While delineating with the assistance of an axial view, make sure to check the other views regularly to see whether it is done correctly.
Figure 10.
The thalamus is gradually losing its oval shape. As the thalamus abuts the midbrain, delineation of the medio-ventral border could become slightly more challenging. Instead of ventricular space, it is now the brachium of the superior colliculus (white matter) briefly defining this border (Mai fig 47, p 186-187). Avoid segmentation of the white matter. Also, while not officially part of the thalamus, the portion of the superior brachium that passes between the pulvinar and the MGN is included in this segmentation because separation is not possible.
Figure 10.
As the lateral geniculate nucleus (LGN) of the thalamus starts to take shape, the thalamus obtains a bulge inferiorly. The capsule of the medial geniculate nucleus (MGN) appears, which, in this image, is separated from the thalamus by the limitans nucleus. Keep avoiding the white matter, but make sure you include the MGN (MAI fig 46, p 184-185) (see figure 10).
Figure 11.
As the MGN is no longer ‘separated’ from the thalamus in the following coronal slices, the inferior border is clearly being formed by both the geniculate nuclei. Aside from the inferior bulge, the inferolateral dent is another key feature here. The pretectal area - situated above the posterior commissure - abuts the thalamus ventromedially. An imaginary line between the caudate and the LGN gives a good indication of the lateral thalamic edge (MAI fig 45, p 182-183). (see figure 11).
Figure 12a.
Figure 12b.
At this point, the medial dorsal nucleus is emerging in addition to the pulvinar (see figure 12a). The lateral ventricle is slowly starting to form the roof of the thalamus. Use a sagittal view for better visualization. Note that more rostrally, the reticular thalamic nucleus (separated from the dorsal thalamus by the external medullary lamina) becomes evident (in figure 12b, above the superior boundary marked in red). Added to the provided sagittal view is an axial perspective for orientation.
Figure 13. Identical coronal slices. The medial dorsal nucleus is in yellow.
The medial dorsal (MD) thalamic nucleus is clearly discernible at this point. For segmentation of the MD, use its border with the other thalamic structures, the internal medullary lamina (white matter), to your advantage (see figure 13).
The third ventricle defines the medial border very well. The reticular nucleus encapsulates the thalamus, with the posterior limb of the internal capsule defining its lateral border. The very first ends of the putamen start to appear (see figure 14). When the putamen first appears varies greatly per individual brain.
Figure 15.
Figure 16.
Figure 17.
Note the LGN gradually ‘separating’ from the rest of the thalamus in coronal slices (see figure 16). At the midline, the habenula (white matter) emerges. To determine its position, check the axial view (see figure 17). To determine the inferomedial border in
coronal sections, it is helpful to imagine a line running from the habenular nucleus in the direction of the (pre)subiculum(Power et al., 2015; MAI f 43, p 178-179).
Figure 18.
Figure 19.
The superior colliculus is no longer visible. The lateral ventricle marks the superior border of the thalamus. The body of the fornix can be seen running through the lateral ventricle. (figure 19).
Figure 20.
In the pons, the substantia nigra and cerebral peduncle are becoming apparent (see figure 20).
Figure 22.
While the habenula has become less visible, the red nucleus (RN) is now evident, at about halfway through the length of thalamus. Positioned ventromedially to the thalamus, the RN can be identified by its rounded structure. It lies dorsomedial to the substantia nigra and cerebral peduncle, as is shown in figure 22. Throughout the following slices, the red nucleus serves as a useful landmark for the definition of the inferior border. On coronal slices, the LGN is not connected to the large thalamic mass any longer. The MGN is no longer visible.
Figure 23.
The cerebral peduncle/internal capsule clearly separates the lateral geniculate nucleus from the rest of the thalamus (see figure 23; MAI fig 39, p 170 – 171). Keep checking axial views for the definition of the lateral thalamic border.
Figure 24.
Keep using the roof of the red nucleus for the definition of the inferior thalamic border in coronal slices (MAI fig 38, p 168 – 169).
Figure 25.
The LGN is no longer in sight. In the next number of slices, the thalamus will keep the same heart-like shape with the other half. The consistency in shape will make boundary definition relatively easy.
Figure 26.
The putamen is clearly visible at this level (see figure 26). The optic tract can be seen close to where the LGN was positioned in earlier slices.
Figure 27.
The thalamus is surrounded by white matter at its lateral and inferior borders. As the two thalami slowly start to meet at the midline, their characteristic heart-shape becomes more evident.
Figure 28.
The rostral end of the external globus pallidus (GP) is appearing medial to the putamen (see figure 28)(MAI fig 37, p 166 – 167).
Figure 30.
The thalamus retains a half-heart shape throughout these slices, while gradually decreasing in size.
Figure 31.
Medial to the external GP and superior to the optic tract, the internal GP is emerging (see figure 31; MAI fig 35, p 162 – 163). This serves as a useful landmark in that it indicates the (near) caudal end of the red nucleus. The zona incerta (ZI) and the subthalamic nucleus (STN) will mostly define the inferior border from thereon.
Figure 32.
At this point, lateral to the MD, the thalamus mostly consists of nuclei of the ventral lateral group. As the RN is no longer visible, the STN (superior to the substantia nigra) and the zona incerta should now be used to define the inferior border of the thalamus. The ZI is contiguous with the reticular nucleus and lies superior to the STN. Make use of the white matter fibers running along the superior length of ZI, as these form a good contrast with the gray matter of the ventral lateral nucleus (see figure 32; MAI fig 35, p 160 – 161).
Figure 34.
While its distinction may vary greatly per individual brain, the mamillo-thalamic tract can be seen running through the ventral anterior thalamic nucleus in a small number of succeeding coronal slices (see figure 34; MAI fig. 32, p 156 – 157).
Figure 38.
Once the STN is starting to disappear and no longer serves as a feasible landmark for the inferior border, the hypothalamus becomes contiguous with the ZI, which will continue to assist with delineation. The (posterior) hypothalamic area lies directly next to the third ventricle, at the same height, and can thus be spotted relatively easily (see figure 36, 37 & 38; MAI fig 31, p 154 - 155).
Figure 41.
The thalamus is becoming smaller, receding superiorly towards the lateral ventricle. The mamillary bodies, inferior to the hypothalamus and medial to the optic tract, are clearly visible and will remain so in the following slices. These are useful for orientation.
Figure 46.
The ventral anterior nucleus (VA) and the anterior thalamic nucleus comprise most of the thalamus at this point, with the internal medullary lamina crossing these two areas. The subthalamic nucleus is no longer visible. The white matter of the internal capsule still marks the inferolateral border well, however.
Figure 50.
The mamillary bodies are becoming contiguous with the hypothalamus. Whereas the lateral and third ventricle still make delineation of the medial and superior border simple, the lateral and inferior border are becoming vaguer. The caudate is increasing in size, and its medial side acts as an estimation of where the lateral edge of the thalamus is. This is an arbitrary indication, and thus zooming in might be necessary to pinpoint the exact lateral thalamic boundary.
Figure 51.
In these last slices, the thalamus decreases in size quickly. The superolateral end of the thalamus (nearly) touches the inferomedial tip of the caudate nucleus. The inferior tip of the thalamus touches the hypothalamus superiorly. Thus, the size of the thalamus can be determined by following the inferomedial end of the caudate to the superior tip of the hypothalamus.
Figure 52.
The size of the thalamus should be roughly the same as that of the caudate at this point. The optic tract is closing in on the midline and the mamillary bodies are not visible anymore. While not always immediately visible, the appearance of the amygdala in coronal slices often gives an indication of the rostral end of the thalamus.
Figure 53.
Figure 54.
The very rostral end of the thalamus is hard to identify. To find and define it, a sagittal view should once more be used in combination with a coronal view, as shown in figure 53 and 54. As was the cast in previous slices, it is helpful to use the inferomedial end of the caudate for orientation. Once the end has been reached, do a last thorough check for discrepancies in all views, focussing on the inferior border in sagittal slices and the lateral border in an axial perspective.
