JOURNAL OF MAGNETIC RESONANCE IMAGING 39:1018–1026 (2014)
Technical Note
High-Resolution MRI and Diffusion-Weighted Imaging of the Human Habenula at 7 Tesla Barbara Strotmann, MS, MS, cand. MD,1 Robin M. Heidemann, PhD,1,2 Alfred Anwander, PhD,3 Marcel Weiss, MS,1 Robert Trampel, PhD,1 Arno Villringer, MD,4 and Robert Turner, PhD1* Key Words: habenula; 7T; structural MRI; diffusionweighted imaging J. Magn. Reson. Imaging 2014;39:1018–1026. C 2013 Wiley Periodicals, Inc. V
Purpose: To investigate the feasibility of discriminating the habenula in human brain using high-resolution structural MRI and diffusion-weighted imaging at 7 Tesla (T). Materials and Methods: MRI experiments included a MP2RAGE and GRE sequence to acquire quantitative parameter maps of T1, T2*, and a calculated proton density map and the combined approach of zoomed and parallel imaging (ZOOPPA) to obtain dw images. Probabilistic tractography algorithms were used to identify multiple fiber orientations in submillimetre voxels, and constrained spherical deconvolution to resolve orientations in regions where fibers cross.
THE HABENULA IS a small structure of approximately 5–9 mm in diameter close to the midline in human brain (Fig. 1). Due to its small size the habenula has been difficult to investigate, and only a few studies have attempted to elucidate its functional role. However, it is becoming apparent that this small structure plays a crucial role in controlling the human reward system, and its dysfunction may underlie several psychiatric disorders. A high spatial resolution and a good contrast to noise ratio (CNR) are required to visualize the habenula in vivo. This can be achieved with MRI at ultra-high field strength (7 Tesla [T] and above). To benefit from the inherently higher signal-to-noise ratio (SNR) at this field strength, new imaging strategies have been used and protocols with optimized sequence parameters have been developed. We explored the human habenula in vivo and ex vivo using these advanced MRI techniques.
Results: Maps of T1, T2*, and proton density showed high contrast of the human habenula. The lateral habenula and its commissure can be distinguished from medial habenula and adjacent tissue. DWI data with 0.7 mm isotropic resolution revealed that fiber populations differ in medial and lateral habenula and two major fiber bundles that connect habenular nuclei with forebrain structures and brainstem. Conclusion: High resolution 7T MR imaging of the human habenula provides sufficient signal-to-noise and contrast to enable identification of the lateral and medial nuclei. In vivo high resolution DWI at 7T is able to distinguish between lateral and medial habenula, and to detect major fiber tracts that connect the habenula with other brain areas.
Habenular Anatomy
1 Department of Neurophysics, Max-Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany. 2 Siemens Healthcare Sector, Erlangen, Germany. 3 Department of Cortical Netwarks, Max-Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany. 4 Department of Cognitive Neurology, Max-Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany. *Address reprint requests to: R. Turner, Department of Neurophysics. Max Planck Institute for Human Cognitive and Brain Sciences. Stephanstrabe 1a, 04103 Leipzig, Germany. E-mail: [email protected] Received January 21, 2013; Accepted May 9, 2013. DOI 10.1002/jmri.24252 View this article online at wileyonlinelibrary.com. C 2013 Wiley Periodicals, Inc. V
The small rein, the habenula, is located above the thalamus next to the third ventricle. Together with the pineal body it is regarded as the epi-thalamus (Figs. 1 and 2). Its commissure, the habenular commissure (hbc), connects the habenular nuclei on both hemispheres forming a trigone in front of the posterior commissure. Histological studies (1,2) divide the habenula into a medial and lateral part with connections to limbic forebrain by means of stria medullaris and brainstem areas by means of fasciculus retroflexus (Fig. 3) (3,4). Medial habenula (MHB) gets mainly input from the limbic system and sends efferents to interpenduncular nuclei (IP) projecting to serotonergic cell regions, medial and dorsal raphe (DR, MR), whereas lateral habenula (LHB) receives mainly input from the basal ganglia and projects mainly to paramedian regions with dopaminergic cell populations, such as
1018
MRI and DWI of the Human Habenula
1019
(6,7). Lateral habenula is over-activated in major depressive disorder which can be influenced and treated by deep brain stimulation (DBS) (8–10). MR Imaging of the Habenula
Figure 1. The habenula is a small (5–9 mm in diameter) but crucial node in the brain next to the third ventricle (as in 27).
substantia nigra pars compacta (SNpc) and ventral tegmental area (VTA), partly through the rostromedial tegmental nucleus (RMTg) (5). The habenula has a control function within the human reward system: positive reward is signaled by the dopamine system, whereas disappointment appears to be linked to increased habenular activity
As mentioned above, the habenula is approximately 5– 9 mm across. To investigate the habenula with structural MRI and DWI, an isotropic resolution in the millimeter or even better in the sub-millimeter regime is necessary. This provides a challenge for in vivo neuroimaging. Because the SNR depends linearly on the image voxel volume, increasing the spatial resolution results in a decreased SNR. An increased tissue contrast (to increase the CNR) based on T1, T2, or both can be achieved by various combinations of inversion times, flip angles, echo times, and sequence repeat times. However, this often entails a concomitant loss in SNR. Little has been published regarding in vivo visualization of the habenula in human brain (11–15). At 3T, attempts to image the anatomy of the human habenula in vivo, and to clearly delineate boundaries with surrounding regions as they vary across
Figure 2. The habenula, the small rein, sits next to the third ventricle above the thalamus before the posterior commissure. Together with the pineal body, the habenula is regarded as the epithalamus. The habenular commissure connects the habenula on both hemispheres and forms a trigone in front of the posterior commissure.
Figure 3. Connectivity of the habenula: The human habenula with its medial (MHb) and lateral nuclei (LHb) connects by means of the stria medullaris to the forebrain and by means of the fasciculus retroflexus toward midbrain regions.
1020
subjects, have met with limited success (16). The higher resolution required has entailed very long scan times and also a loss of SNR. Consequently, the contrast-to-noise ratio is insufficient to distinguish and clearly identify the existence of subdivisions, nor to discriminate unambiguously the functional activity considered to be associated with the lateral habenula. Ultra high field MRI offers increased inherent SNR, roughly proportional to the strength of the main magnetic field. Contrast relating to the iron content of tissue is disproportionately improved with increasing field strength. Due to these potential benefits, scanning was therefore performed at 7T, providing the high resolution and high contrast necessary to investigate the habenula in vivo. Diffusion-weighted MRI (17,18) has been used for more than 15 years as a guide to axonal organization in white matter. DWI can now be performed with isotropic sub-millimeter resolution, enabling study of more complex fiber architectures (19). Recently, several studies have focused on structures such as the amygdala, thalamus, mid-brain, and brain stem (eg, (20–23)). However, the axonal connectivity of the habenula has never been explored noninvasively. The structure, function, and connectivity of the habenula of the human brain are thus not yet well understood, and have not been fully investigated using MRI and DWI in vivo. In particular, a better understanding of habenular subdivisions and their specific roles in brain function is needed. The aim of the present study is to identify the anatomic details observable in high-resolution MR images and DWI data of the human habenula, comparing in vivo and ex vivo findings, and relating these to established histological maps.
Strotmann et al.
ms, TE1 ¼ 8.2 ms, TE2 ¼ 15.3 ms, TE3 ¼ 22.43 ms, TE4 ¼ 29.57 ms, BW ¼ 200 Hz/px, 0.8 mm isotropic resolution, flip angle ¼ 15 ). For the determination of the regional absolute water content and evaluation of dynamics in our region of interest, we then calculated a map of the density of protons in the imaged volume using an exponential fit (S ¼ S0 exp [TE/T2*]) (25,26). The maps were co-registered, and the habenula was identified by visual inspection and comparison with surrounding macroanatomical landmarks (27,28). Ex Vivo Imaging To investigate the characteristics of human habenula we compared the in vivo structural MR scans with ex vivo MR anatomy. To this end, a post mortem brain was fixed in 4% formalin 24 h after death (female, 65 years old, cardiac failure) (Fig. 2). T1 maps were acquired using the MP2RAGE sequence (TR ¼ 3000 ms, TE ¼ 7.38 ms, BW ¼ 170 Hz/px, 0.15 mm isotropic re solution, flip angle ¼ 8 ) and maps of T2* were acquired using a spoiled three-dimensional (3D) gradient-echo sequence (TR ¼ 35 ms, TE1 ¼ 7.7 ms, TE2 ¼ 15.4 ms, TE3 ¼ 23.0 ms, 0.25 mm isotropic resolution, flip angle ¼ 10 ). Additionally, to minimize the effects of T1 and T2, we used a 3D gradient-echo sequence with a range of echo times (TR ¼ 1000 ms, TE1 ¼ 6.3 ms, TE2 ¼ 16.68 ms, TE3 ¼ 27.06 ms, TE4 ¼ 37.44 ms, TE5 ¼ 47.82 ms, TE6 ¼ 58.20 ms, TE7 ¼ 68.58 ms, TE8 ¼ 78.96 ms, 0.25 mm isotropic resolution, flip angle ¼ 68.4 ) and calculated proton density maps accordingly. Diffusion Imaging at 7T
MATERIALS AND METHODS Three subjects were scanned on a 7T whole-body MR scanner using a 24-channel phased-array head coil (Nova Medical Inc, Wilmington, MA). The study was approved by the ethics committee of the local university and informed consent was obtained. The ex vivo samples were scanned using a custom built single channel square coil (120 mm side) and the tissue was centered inside. Structural MR Imaging at 7T In Vivo Imaging high-resolution whole-brain T1 maps were acquired in vivo using the magnetization prepared 2 rapid acquisition gradient echoes sequence (MP2RAGE) (24). This sequence provides a bias-field-free T1-weighted acquisition with improved T1 contrast. Furthermore, quantitative T1-maps can be obtained. For this purpose, the following imaging parameters have been used: (repetition time [TR] ¼ 4520 ms, echo time [TE] ¼ 3.8 ms, TI1 ¼ 1100 ms, TI2 ¼ 3500 ms, bandwidth [BW] ¼ 170 Hz/px, 0.8 mm isotropic resolution, flip angle 1 ¼ 4, flip angle 2 ¼ 4 ). Maps of T2* were acquired using a spoiled gradientecho sequence with a range of echo times (TR ¼ 49
To allow a precise localization of the region of interest, we first acquired a 3D T1-weighted data set using the MP2RAGE sequence (TR ¼ 6000 ms, TI1 ¼ 900 ms, TI2 ¼ 2750 ms, TE ¼ 3.11 ms, 0.7 mm isotropic resolu tion, flip angle 1 and 2 ¼ 4 and 3 ). DW images were acquired with optimized Stejskal-Tanner diffusion weighting scheme (29) and a novel technique, a combined approach of reduced FOV acquisition (zoomed imaging) and parallel imaging (GRAPPA), given the name ZOOPPA (zoomed partially parallel acquisitions), was applied (19). For high-resolution diffusion MRI at ultra high field strength, it is necessary to address the problems of susceptibility artifacts and image blurring. This can be achieved by accelerating the acquisition using parallel imaging (30). However, for this high resolution and field strength, it is necessary to achieve high acceleration factors with a minimum amount of additional noise, arising from the parallel image reconstruction. A combination of zoomed imaging (31,32) and parallel imaging as described by Heidemann et al (33) can be used to significantly improve the image quality of single-shot EPI acquisitions. This method is beneficial for high resolution DWI at ultra high field strength (19). We used a protocol with 1 mm isotropic resolution with the following parameters: TR ¼ 10400 ms, TE ¼ 82 ms, FOV ¼ 144 150 mm2, image matrix 144 150, 75%
MRI and DWI of the Human Habenula
1021
Figure 4. Habenula in vivo (axial view): T1, T2*, and proton density maps show the human habenula in vivo. HB, habenula; AC, anterior commissure; PC, posterior commissure.
partial Fourier, total acceleration factor 4.2. Fat suppression was performed with a combination of the gradient reversal method (34) and a low SAR approach (35) as described in Eichner et al (36). Diffusion weighting was performed along 60 diffusionencoding gradient directions with a b-value of 1000 s/mm2. Seven images with no diffusion weighting were placed at the beginning of the sequence and after each block of 10 diffusion-weighted images, to provide anatomical reference for offline motion correction. The interleaved measurement of 71 slices with 10% overlap covered most of the brain in the anterior–posterior direction. Four averages were acquired, resulting in a total acquisition time of 48 min for 71 slices.
software package (Brain Research Institute, Melbourne, Australia, www.brain.org.au/software/). In a separate analysis, multiple orientations of crossing fibers were modeled using constrained harmonized spherical deconvolution (SCD) (41,42) and probabilistic streamline tractography methods were applied. With ultra-high resolution data more fiber orientation distribution (FOD) parameters have to be estimated than actually measured. For this reason, spherical harmonized deconvolution can resolve orientations with smaller angles, and provide a robust estimation of the distribution of fiber orientations in each voxel while preserving angular resolution.
RESULTS Data Processing
In Vivo Imaging at 7T
The DW images were noise cleaned using a two-stage hybrid image restoration procedure (37). In each series, the seven images without diffusion weighting were used to estimate motion correction parameters using rigid-body transformations (38) as implemented in FSL (FMRIB Software Library, University of Oxford 2006, www.fmrib.ox.ac.uk/fsl). The motion correction for the 60 diffusion-weighted images was combined with global registration to the T1 anatomy computed by the same method. The gradient direction for each volume was corrected using the rotation parameters. The registered images were interpolated to the new reference frame, a quantitative T1-weighted anatomical scan with an isotropic voxel resolution of 0.7 mm.
In each voxel, multiple fibers orientations were computed using the ball-and-2-sticks model (39,40). Whole-brain tracking was performed using the MRtrix
Figure 4 shows, in axial sections through the habenula, computed parameter maps of the relaxation times T1, T2*, and proton density (PD). Directly below the ventricular surface, this nucleus can be clearly visualized on both T1 and T2* maps, medial to the caudal part of the dorsal thalamus. The prominent structure shows signal characteristics of both gray matter and white matter. T1 images show a strong contrast of the human habenula with surrounding brain tissue that would typically result from a high myelin content. Probably due to a concomitant high iron content, the habenula also shows reduced T2* compared with neighboring regions. In addition, proton density data from measurements of the regional absolute water content in vivo reveal high contrast between the habenula and adjacent brain. Overall, we can distinguish the human habenula in vivo from surrounding brain tissue based on three different MR contrasts. Table 1 shows the values in human habenula on quantitative parameter maps.
Table 1 Values of the Human Habenula on Quantitative Parameter Maps
Ex Vivo Imaging at 7T
Fiber Tracking
T1 (in ms)
T2* (in ms)
1312–1500 (max. 1900)
28–38
PD (a.u.) 1035–1120 (max. 1335)
Figure 5 illustrates the human habenula ex vivo in the axial plane on maps of T1, T2*, and proton density (PD). Histological studies define the lateral habenula
1022
Strotmann et al.
Figure 5. Habenula ex vivo (axial view): T1, T2*, and proton density maps show distinct lateral and medial habenular nuclei according to myelin stain sections (adapted from 43). MHB, medial habenula; LHB, lateral habenula; hbc, habenular commissure; pc, posterior commissure.
(LHB) and its commissure (hbc) by higher myelin content (included in Fig. 5, of Riley [43]). On axial MR slices, subdivisions of the human habenula become visible. In both T1 and T2* maps the medial and lateral nuclei of the habenula can be distinguished. The Lateral Habenula (LHB) and the habenular commissure (HBC) show a high contrast from surrounding brain and clearly differ from medial habenula (MHB). Overall, in these ex vivo images at 7T, the anatomy of the human habenula can be clearly depicted, and its major subdivisions distinguished. Diffusion Imaging at 7T A probabilistic multi-fiber fit to the 60 direction data set supports more than one fiber orientation in many voxels (Fig. 6). Note the color-coded main fiber orientations with significant signal contribution (threshold: f-value > 0.05) in both the axial plane and, more notably, the coronal plane. Brain sections containing the habenula were superimposed on the quantitative high-resolution T1 map
and compared with histology of stained sections (43,44) and Mai Atlas (27). T1 images show strong contrast between the lateral habenula and the habenular commissure with the surrounding brain tissue, whereas medial habenula shows less contrast. However, both medial and lateral habenula can be clearly seen in the tractography maps. We can distinguish medial habenular nuclei (MBH), having only one fiber orientation in the anterior–posterior direction (green), while lateral habenula (LHB) shows two orientations, anterior–posterior and also superior–inferior (blue), as revealed on the horizontal view. The habenular commissure (HBC) displays a right–left orientation (red), connecting both hemispheres. On coronal view a high level of anatomic detail with corresponding anatomical regions of Mai Atlas (Fig. 7) reveals subdivisions of lateral habenula: fiber populations in LHB do not contribute similarly to the MR signal. Combining T1 contrast and diffusion properties we observe a border region between LHB and MHB: there is a lateral part of LHB with a high myelin content (l-LHB) and a medial part with different diffusion
MRI and DWI of the Human Habenula
1023
Figure 6. Coronal (top) and horizontal (bottom) view of the habenula diffusion with the colorcoded vector orientations showing lateral habenula (LHB), medial (MHB), and the habenular commissure (HBC), and the way of stria medullaris (STM).
properties close to MHB marked with * containing less myelin. Probabilistic tractography shows the two major fiber bundles that connect habenular nuclei with limbic forebrain structures and brainstem on a horizontal slice (Fig. 8). Fibers from LHB appear to pass mainly to the forebrain, while tracts from MHB go down to the brainstem and to its commissure. This is illustrated by streamline tractography using spherical deconvolution. While fibers from LHB seem to run mainly to the forebrain, tracts from MHB go down to the brainstem and to its commissure as illustrated by streamline tractography using spherical deconvolution (Fig. 9). Descending fibers can be detected running from the stria medullaris (STM) to forebrain regions and projections caudally by the way of the fasciculus retroflexus (FR) to paramedian midbrain regions.
DISCUSSION MR imaging at 7T provides sufficient spatial resolution to reveal anatomical details of the human habenula. At ultra-high field strength, there is enough SNR to allow 0.4 mm isotropic voxels in vivo (45–47) for anatomical MRI and MR-Microscopy of ex vivo
brain samples. Using the high field strength of 7T and a combination of zoomed imaging with parallel imaging (ZOOPPA) enables DWI acquisitions with 800 mm to 1 mm isotropic resolution (19). Our data show a high SNR, allowing excellent brain tissue differentiation. Using high performance gradients at 7T, we can resolve fiber crossings at submillimeter resolution and visually identify the human habenula and its substructures in vivo. With spherical harmonic deconvolution and robust probabilistic tractography, complex fiber architecture is revealed in this small region. The internal anatomy of this brain organelle has not previously been investigated using MR techniques. Most of the evidence for the structure and functions of the habenula has been drawn from rodent research, using mainly histological techniques. For the first time, structural MRI reveals the major basic units of the habenular complex: the lateral habenula, the medial habenula, and the habenular commissure. Now, in vivo DWI data discriminate the lateral habenula, which can be clearly distinguished from adjacent brain tissue, and show the main fiber bundles going through this node. The human habenula shows good contrast on both T1 and T2* parameter maps. This suggests strongly that it is characterized by a high concentration both
1024
Strotmann et al.
Figure 7. Coronal view of the habenula diffusion with the color coded vector-orientation zoomed in compared with Mai Atlas (26) showing correspondence. LHb, lateral habenula; MHb, medial habenula; Lim, nucleus limitans; PTc: pretectal area; pc, posterior commissure; Aq, aqueduct; PAG, periaqueductal gray; MD, medial dorsal thalamic nucleus; iml, internal medullary lamina of thalamus, and * a region making border between LHb and MHb.
of myelin (T1 decrease) and of iron (T2* decrease). Ex vivo, the lateral human habenula shows particularly high contrast with surrounding brain tissue on maps of T1 and T2* contrast, as well as on spin density maps. This enables parcellation of both medial and lateral habenular nuclei. Whereas in healthy brain tissue T1 contrast largely depends on myelin, there are several mechanisms
Figure 8. Identification of fiber tracts leaving the habenula. STM, stria medullaris; FR, fasciculus retroflexus.
that can affect T2*. Recent research suggests that increased tissue iron content, in the form of ferretin and/or neuromelanin, is often the reason for a decrease in T2* (48,49). Furthermore, using data from the MP2RAGE sequence, which compensates for B1 field inhomogeneities and tissue heterogeneity, we were able to analyze proton density contrast. Enhanced proton density in the human habenula suggests increased water content. In conclusion, 7T MRI provides good contrast and clear identification of the habenula in vivo, and shows striking internal details in cadaver human brain samples. It is unlikely that similar results could be achieved at lower field strengths. At 7T, we can distinguish lateral, medial habenula, and the habenular commissure. The combined approach of zoomed imaging and parallel imaging (ZOOPPA) was used to acquire diffusion weighted MRI (DWI), which provided exceptionally high resolution, and greatly reduced problems of susceptibility artifacts and image blurring. DWI at 7T can show mainly descending fibers from the habenula, the stria medullaris to forebrain regions and caudally directed projections, in the form of the fasciculus retroflexus to paramedian midbrain
MRI and DWI of the Human Habenula
Figure 9. Corresponding fiber tractography with seed points in lateral habenula (LHB), medial habenula (MHB), and the habenular commissure (HBC).
regions. Overall, we can distinguish lateral and medial habenula, and its commissure, based on high resolution DWI data in vivo. Previously had been mainly reported projections of LHB from basal ganglia to dopaminergic cell regions and limbic afferents of MHB and efferents to serotonergic cell regions and interpeduncular nucleus. LHB showed projections to forebrain regions, whereas MHB showed mainly descending fibers and projections to the habenular commissure. Histology of rat and cat brain shows a heterogeneous lateral habenula (1,50,51): the lateral part (l-LHB) is assumed to be linked to the motor system and the medial part to the limbic system (m-LHB). Future research is planned to investigate structural connectivity and functional activation at high spatial resolution in the human habenula, which may assist in the investigation of the pathophysiology of a wide range of neurologic and psychiatric disorders. In conclusion, further studies of the habenular subdivisions and their role in brain function is likely to improve its understanding and might become crucial for the placement of electrodes for deep brain stimulation in major depression (10). Studies of habenula contrast, as it reflects iron and myelin content, and of its subdivisions, are the subject of further research, which might assist in the understanding of health and clinical disorders. ACKNOWLEDGEMENTS We thank Domenica Wilfing and Elisabeth Wladimirow for MRI scanning and subject recruitment, and Erik T€ urke and Enrico Reimer for technical support. REFERENCES 1. Herkenham M, Nauta WJ. Afferent connections of the habenular nuclei in the rat. A horseradish peroxidase study, with a note on the fiber-of-passage problem. J Comp Neurol 1977;173:123–146. 2. Herkenham M, Nauta WJ. Efferent connections of the habenular nuclei in the rat. J Comp Neurol 1979;187:19–47. 3. Sutherland RJ. The dorsal diencephalic conduction system: a review of the anatomy and functions of the habenular complex. Neurosci biobehav Rev 1982;6:1–13. 4. Hikosaka O, Sesack SR, Lecourtier L, et al. Habenula: crossroad between the basal ganglia and the limbic system. J Neurosci 2008;28:11825–11829.
1025 5. Balcita-Pedicino J, Sesack SR. Ultrastructural analysis of the projection from the lateral hypothalamic area to the lateral habenula in the rat. Soc Neurosci Abstr 2010;36:511–530. 6. Matsumoto M, Hikosaka O. Lateral habenula as a source of negative reward signals in dopamine neurons. Nature 2007;447: 1111–1115. 7. Hikosaka O. The habenula: from stress evasion to value-based decision-making. Nat Rev Neurosci 2010;11:503–513. 8. Morris JS, Smith KA, Cowen PJ, et al. Covariation of activity in habenula and dorsal raphe nuclei following tryptophan depletion. Neuroimage 1999;10:163–172. 9. Sartorius A, Henn FA. Deep brain stimulation of the lateral habenula in treatment resistant major depression. Med Hypotheses 2007;69:1305–1308. 10. Sartorius A, Kiening KL, Kirsch P, et al. Remission of major depression under deep brain stimulation of the lateral habenula in a therapy-refractory patient. Biol Psychiatry 2010;67:e9–e11. 11. Ullsperger M, von Cramon DY. Error monitoring using external feedback: specific roles of the habenular complex, the reward system, and the cingulate motor area revealed by functional magnetic resonance imaging. J Neurosci 2003;23:4308–4314. 12. Salas R, Baldwin P, de Biasi M, et al. BOLD responses to negative reward prediction errors in human habenula. Front Hum Neurosci 2010;4:36. 13. Ide JS, Li CS. Error-related functional connectivity of the habenula in humans. Front Hum Neurosci 2011;5:25. 14. Shelton L, Pendse G, Maleki N, et al. Mapping pain activation and connectivity of the human habenula. J Neurophysiol 2012; 107:2633–2648. 15. Lawson RP, Drevets WC, Roiser JP. Defining the habenula in human neuroimaging studies. Neuroimage 2013:722–727. 16. Savitz JB, Nugent AC, Bogers W, et al. Habenula volume in bipolar disorder and major depressive disorder: a high-resolution magnetic resonance imaging study. Biol Psychiatry 2011;69:336–343. 17. Turner R, Le Bihan D. Single-shot diffusion imaging at 2.0 Tesla. J Magn Reson 1990;86:445–452. 18. Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson 1994;103:247–254. 19. Heidemann RM, Anwander A, Feiweier T, Knosche TR, Turner R. k-space and q-space: combining ultra-high spatial and angular resolution in diffusion imaging using ZOOPPA at 7 T. Neuroimage 2012;60:967–978. 20. Solano-Castiella E, Anwander A, Lohmann G, et al. Diffusion tensor imaging segments the human amygdala in vivo. Neuroimage 2010;49:4:2958–2965. 21. Chanraud S, Reynaud M, Wessa M, et al. Diffusion tensor tractography in mesencephalic bundles: relation to mental flexibility in detoxified alcohol-dependent subjects. Neuropsychopharmacology 2009;34:5:1223–1232. 22. Yeo SS, Chang MC, Kwon YH, et al. Corticoreticular pathway in the human brain: diffusion tensor tractography study. Neurosci Lett 2012;508:9–12. 23. Yang HS, Kwon HG, Hong JH, et al. The rubrospinal tract in the human brain: diffusion tensor imaging study. Neurosci Lett 2011;504:45–48. 24. Marques JP, Kober T, Krueger G, et al. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. Neuroimage 2010;49:1271–1281. 25. Neeb H, Zilles K, Shah NJ. A new method for fast quantitative mapping of absolute water content in vivo. Neuroimage 2006;31: 1156–1168. 26. Warntjes JB, Dahlqvist O, Lundberg P. Novel method for rapid, simultaneous T1, T*2, and proton density quantification. Magn Reson Med 2007;57:528–537. 27. Mai JK, Voss T, Paxinos G, Voss. Atlas of the human brain, 3rd edition. Boston: Elsevier; 2008. p 180–182. 28. Nieuwenhuys R, Voogd J, Van Huijzen C. The human central nervous system. Heidelberg: Springer; 2008. p 79–81. 29. Morelli JN, Runge VM, Feiweier T, et al. Evaluation of a modified Stejskal-Tanner diffusion encoding scheme, permitting a marked reduction in TE, in diffusion-weighted imaging of stroke patients at 3 T. Invest Radiol 2011;45:29–35. 30. Griswold MA, Jakob PM, Chen Q, et al. Resolution enhancement in single-shot imaging using simultaneous acquisition of spatial harmonics (SMASH). Magn Reson Med 1999;41:1236–1245.
1026 31. Feinberg DJ, Hoenninger J, Crooks L, et al. Inner volume MR imaging: technical concepts and their application. Radiology 1985;156:743–747. 32. Mansfield P, Ordidge RJ, Coxon R. Zonally magnified EPI in real time by NMR. J Phys E 1988:275–280. 33. Heidemann RM, Ivanov D, Trampel R, et al. Isotropic sub-millimeter fMRI in the human brain at 7 Tesla: Combining Reduced Field-of-View. Magn Reson Med 2012; 68(5): 1506–1516. 34. Nagy Z, Weiskopf N. Efficient fat suppression by slice-selection gradient reversal in twice-refocused diffusion encoding. Magn Reson Med 2008;60:1256–1260. 35. Ivanov D, Schaefer A, Streicher MN, et al. A simple low-SAR technique for chemical-shift selection with high-field spin-echo imaging. Magn Reson Med 2010;64:319–326. 36. Eichner C, Ivanov D, Feiweier T, et al. High resolution diffusionweighted imaging at 7T - issues of slice thickness and fat suppression. Poster presented at 17th Annual Meeting of the Organization for Human Brain Mapping, Quebec City, QC, Canada, 2011. (abstract 560). 37. Lohmann G, Bohn S, Muller K, et al. Image restoration and spatial resolution in 7-tesla magnetic resonance imaging. Magn Reson Med 2010;64:15–22. 38. Jenkinson M, Bannister P, Brady M, et al. Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 2002;17:825–841. 39. Behrens TE, Berg HJ, Jbabdi S, et al. Probabilistic diffusion tractography with multiple fibre orientations: what can we gain? Neuroimage 2007;34:144–155. 40. Behrens TE, Woolrich MW, Jenkinson M, et al. Characterization and propagation of uncertainty in diffusion-weighted MR imaging. Magn Reson Med 2003;50:1077–1088.
Strotmann et al. 41. Tournier JD, Calamante F, Gadian DG, et al. Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution. Neuroimage 2004;23: 1176–1185. 42. Tournier JD, Calamante F, Connelly A. Robust determination of the fibre orientation distribution in diffusion MRI: non-negativity constrained super-resolved spherical deconvolution. Neuroimage 2007;35:1459–1472. 43. Riley H. (1943). An atlas of the basal ganglia, brain stem and spinal cord. Baltimore, MD: Williams & Wilkins; p 427. 44. Ranft K, Dobrowolny H, Krell D, et al. Evidence for structural abnormalities of the human habenular complex in affective disorders but not in schizophrenia. Psychol Med 2010;40:557–567. 45. Collins CM, Smith MB. Calculations of B(1) distribution, SNR, and SAR for a surface coil adjacent to an anatomically-accurate human body model. Magn Reson Med 2001;45:692–699. 46. Ladd ME. High-field-strength magnetic resonance: potential and limits. Top Magn Reson Imaging 2007;18:139–152. 47. Heidemann RM, Seiberlich N, Griswold MA, et al. Perspectives and limitations of parallel MR imaging at high field strengths. Neuroimaging Clin N Am 2006;16:311–320, xi. 48. Haacke EM, Cheng NY, House MJ, et al. Imaging iron stores in the brain using magnetic resonance imaging. Magn Reson Imaging 2005;23:1–25. 49. Gelman N, Gorell JM, Barker PB, et al. MR imaging of human brain at 3.0 T: preliminary report of transverse relaxation rates and relation to estimated iron content. Radiology 1999;210: 759–767. 50. Andres KH, von D€ uring M, Veh RW. Subnuclear organization of the rat habenular complexes. J Comp Neurol 1999;407:130–150. 51. Iwahori N. A Golgi study on the habenular nucleus of the cat. J Comp Neurol 1977;72:3:319–344.
Technical Note
High-Resolution MRI and Diffusion-Weighted Imaging of the Human Habenula at 7 Tesla Barbara Strotmann, MS, MS, cand. MD,1 Robin M. Heidemann, PhD,1,2 Alfred Anwander, PhD,3 Marcel Weiss, MS,1 Robert Trampel, PhD,1 Arno Villringer, MD,4 and Robert Turner, PhD1* Key Words: habenula; 7T; structural MRI; diffusionweighted imaging J. Magn. Reson. Imaging 2014;39:1018–1026. C 2013 Wiley Periodicals, Inc. V
Purpose: To investigate the feasibility of discriminating the habenula in human brain using high-resolution structural MRI and diffusion-weighted imaging at 7 Tesla (T). Materials and Methods: MRI experiments included a MP2RAGE and GRE sequence to acquire quantitative parameter maps of T1, T2*, and a calculated proton density map and the combined approach of zoomed and parallel imaging (ZOOPPA) to obtain dw images. Probabilistic tractography algorithms were used to identify multiple fiber orientations in submillimetre voxels, and constrained spherical deconvolution to resolve orientations in regions where fibers cross.
THE HABENULA IS a small structure of approximately 5–9 mm in diameter close to the midline in human brain (Fig. 1). Due to its small size the habenula has been difficult to investigate, and only a few studies have attempted to elucidate its functional role. However, it is becoming apparent that this small structure plays a crucial role in controlling the human reward system, and its dysfunction may underlie several psychiatric disorders. A high spatial resolution and a good contrast to noise ratio (CNR) are required to visualize the habenula in vivo. This can be achieved with MRI at ultra-high field strength (7 Tesla [T] and above). To benefit from the inherently higher signal-to-noise ratio (SNR) at this field strength, new imaging strategies have been used and protocols with optimized sequence parameters have been developed. We explored the human habenula in vivo and ex vivo using these advanced MRI techniques.
Results: Maps of T1, T2*, and proton density showed high contrast of the human habenula. The lateral habenula and its commissure can be distinguished from medial habenula and adjacent tissue. DWI data with 0.7 mm isotropic resolution revealed that fiber populations differ in medial and lateral habenula and two major fiber bundles that connect habenular nuclei with forebrain structures and brainstem. Conclusion: High resolution 7T MR imaging of the human habenula provides sufficient signal-to-noise and contrast to enable identification of the lateral and medial nuclei. In vivo high resolution DWI at 7T is able to distinguish between lateral and medial habenula, and to detect major fiber tracts that connect the habenula with other brain areas.
Habenular Anatomy
1 Department of Neurophysics, Max-Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany. 2 Siemens Healthcare Sector, Erlangen, Germany. 3 Department of Cortical Netwarks, Max-Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany. 4 Department of Cognitive Neurology, Max-Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany. *Address reprint requests to: R. Turner, Department of Neurophysics. Max Planck Institute for Human Cognitive and Brain Sciences. Stephanstrabe 1a, 04103 Leipzig, Germany. E-mail: [email protected] Received January 21, 2013; Accepted May 9, 2013. DOI 10.1002/jmri.24252 View this article online at wileyonlinelibrary.com. C 2013 Wiley Periodicals, Inc. V
The small rein, the habenula, is located above the thalamus next to the third ventricle. Together with the pineal body it is regarded as the epi-thalamus (Figs. 1 and 2). Its commissure, the habenular commissure (hbc), connects the habenular nuclei on both hemispheres forming a trigone in front of the posterior commissure. Histological studies (1,2) divide the habenula into a medial and lateral part with connections to limbic forebrain by means of stria medullaris and brainstem areas by means of fasciculus retroflexus (Fig. 3) (3,4). Medial habenula (MHB) gets mainly input from the limbic system and sends efferents to interpenduncular nuclei (IP) projecting to serotonergic cell regions, medial and dorsal raphe (DR, MR), whereas lateral habenula (LHB) receives mainly input from the basal ganglia and projects mainly to paramedian regions with dopaminergic cell populations, such as
1018
MRI and DWI of the Human Habenula
1019
(6,7). Lateral habenula is over-activated in major depressive disorder which can be influenced and treated by deep brain stimulation (DBS) (8–10). MR Imaging of the Habenula
Figure 1. The habenula is a small (5–9 mm in diameter) but crucial node in the brain next to the third ventricle (as in 27).
substantia nigra pars compacta (SNpc) and ventral tegmental area (VTA), partly through the rostromedial tegmental nucleus (RMTg) (5). The habenula has a control function within the human reward system: positive reward is signaled by the dopamine system, whereas disappointment appears to be linked to increased habenular activity
As mentioned above, the habenula is approximately 5– 9 mm across. To investigate the habenula with structural MRI and DWI, an isotropic resolution in the millimeter or even better in the sub-millimeter regime is necessary. This provides a challenge for in vivo neuroimaging. Because the SNR depends linearly on the image voxel volume, increasing the spatial resolution results in a decreased SNR. An increased tissue contrast (to increase the CNR) based on T1, T2, or both can be achieved by various combinations of inversion times, flip angles, echo times, and sequence repeat times. However, this often entails a concomitant loss in SNR. Little has been published regarding in vivo visualization of the habenula in human brain (11–15). At 3T, attempts to image the anatomy of the human habenula in vivo, and to clearly delineate boundaries with surrounding regions as they vary across
Figure 2. The habenula, the small rein, sits next to the third ventricle above the thalamus before the posterior commissure. Together with the pineal body, the habenula is regarded as the epithalamus. The habenular commissure connects the habenula on both hemispheres and forms a trigone in front of the posterior commissure.
Figure 3. Connectivity of the habenula: The human habenula with its medial (MHb) and lateral nuclei (LHb) connects by means of the stria medullaris to the forebrain and by means of the fasciculus retroflexus toward midbrain regions.
1020
subjects, have met with limited success (16). The higher resolution required has entailed very long scan times and also a loss of SNR. Consequently, the contrast-to-noise ratio is insufficient to distinguish and clearly identify the existence of subdivisions, nor to discriminate unambiguously the functional activity considered to be associated with the lateral habenula. Ultra high field MRI offers increased inherent SNR, roughly proportional to the strength of the main magnetic field. Contrast relating to the iron content of tissue is disproportionately improved with increasing field strength. Due to these potential benefits, scanning was therefore performed at 7T, providing the high resolution and high contrast necessary to investigate the habenula in vivo. Diffusion-weighted MRI (17,18) has been used for more than 15 years as a guide to axonal organization in white matter. DWI can now be performed with isotropic sub-millimeter resolution, enabling study of more complex fiber architectures (19). Recently, several studies have focused on structures such as the amygdala, thalamus, mid-brain, and brain stem (eg, (20–23)). However, the axonal connectivity of the habenula has never been explored noninvasively. The structure, function, and connectivity of the habenula of the human brain are thus not yet well understood, and have not been fully investigated using MRI and DWI in vivo. In particular, a better understanding of habenular subdivisions and their specific roles in brain function is needed. The aim of the present study is to identify the anatomic details observable in high-resolution MR images and DWI data of the human habenula, comparing in vivo and ex vivo findings, and relating these to established histological maps.
Strotmann et al.
ms, TE1 ¼ 8.2 ms, TE2 ¼ 15.3 ms, TE3 ¼ 22.43 ms, TE4 ¼ 29.57 ms, BW ¼ 200 Hz/px, 0.8 mm isotropic resolution, flip angle ¼ 15 ). For the determination of the regional absolute water content and evaluation of dynamics in our region of interest, we then calculated a map of the density of protons in the imaged volume using an exponential fit (S ¼ S0 exp [TE/T2*]) (25,26). The maps were co-registered, and the habenula was identified by visual inspection and comparison with surrounding macroanatomical landmarks (27,28). Ex Vivo Imaging To investigate the characteristics of human habenula we compared the in vivo structural MR scans with ex vivo MR anatomy. To this end, a post mortem brain was fixed in 4% formalin 24 h after death (female, 65 years old, cardiac failure) (Fig. 2). T1 maps were acquired using the MP2RAGE sequence (TR ¼ 3000 ms, TE ¼ 7.38 ms, BW ¼ 170 Hz/px, 0.15 mm isotropic re solution, flip angle ¼ 8 ) and maps of T2* were acquired using a spoiled three-dimensional (3D) gradient-echo sequence (TR ¼ 35 ms, TE1 ¼ 7.7 ms, TE2 ¼ 15.4 ms, TE3 ¼ 23.0 ms, 0.25 mm isotropic resolution, flip angle ¼ 10 ). Additionally, to minimize the effects of T1 and T2, we used a 3D gradient-echo sequence with a range of echo times (TR ¼ 1000 ms, TE1 ¼ 6.3 ms, TE2 ¼ 16.68 ms, TE3 ¼ 27.06 ms, TE4 ¼ 37.44 ms, TE5 ¼ 47.82 ms, TE6 ¼ 58.20 ms, TE7 ¼ 68.58 ms, TE8 ¼ 78.96 ms, 0.25 mm isotropic resolution, flip angle ¼ 68.4 ) and calculated proton density maps accordingly. Diffusion Imaging at 7T
MATERIALS AND METHODS Three subjects were scanned on a 7T whole-body MR scanner using a 24-channel phased-array head coil (Nova Medical Inc, Wilmington, MA). The study was approved by the ethics committee of the local university and informed consent was obtained. The ex vivo samples were scanned using a custom built single channel square coil (120 mm side) and the tissue was centered inside. Structural MR Imaging at 7T In Vivo Imaging high-resolution whole-brain T1 maps were acquired in vivo using the magnetization prepared 2 rapid acquisition gradient echoes sequence (MP2RAGE) (24). This sequence provides a bias-field-free T1-weighted acquisition with improved T1 contrast. Furthermore, quantitative T1-maps can be obtained. For this purpose, the following imaging parameters have been used: (repetition time [TR] ¼ 4520 ms, echo time [TE] ¼ 3.8 ms, TI1 ¼ 1100 ms, TI2 ¼ 3500 ms, bandwidth [BW] ¼ 170 Hz/px, 0.8 mm isotropic resolution, flip angle 1 ¼ 4, flip angle 2 ¼ 4 ). Maps of T2* were acquired using a spoiled gradientecho sequence with a range of echo times (TR ¼ 49
To allow a precise localization of the region of interest, we first acquired a 3D T1-weighted data set using the MP2RAGE sequence (TR ¼ 6000 ms, TI1 ¼ 900 ms, TI2 ¼ 2750 ms, TE ¼ 3.11 ms, 0.7 mm isotropic resolu tion, flip angle 1 and 2 ¼ 4 and 3 ). DW images were acquired with optimized Stejskal-Tanner diffusion weighting scheme (29) and a novel technique, a combined approach of reduced FOV acquisition (zoomed imaging) and parallel imaging (GRAPPA), given the name ZOOPPA (zoomed partially parallel acquisitions), was applied (19). For high-resolution diffusion MRI at ultra high field strength, it is necessary to address the problems of susceptibility artifacts and image blurring. This can be achieved by accelerating the acquisition using parallel imaging (30). However, for this high resolution and field strength, it is necessary to achieve high acceleration factors with a minimum amount of additional noise, arising from the parallel image reconstruction. A combination of zoomed imaging (31,32) and parallel imaging as described by Heidemann et al (33) can be used to significantly improve the image quality of single-shot EPI acquisitions. This method is beneficial for high resolution DWI at ultra high field strength (19). We used a protocol with 1 mm isotropic resolution with the following parameters: TR ¼ 10400 ms, TE ¼ 82 ms, FOV ¼ 144 150 mm2, image matrix 144 150, 75%
MRI and DWI of the Human Habenula
1021
Figure 4. Habenula in vivo (axial view): T1, T2*, and proton density maps show the human habenula in vivo. HB, habenula; AC, anterior commissure; PC, posterior commissure.
partial Fourier, total acceleration factor 4.2. Fat suppression was performed with a combination of the gradient reversal method (34) and a low SAR approach (35) as described in Eichner et al (36). Diffusion weighting was performed along 60 diffusionencoding gradient directions with a b-value of 1000 s/mm2. Seven images with no diffusion weighting were placed at the beginning of the sequence and after each block of 10 diffusion-weighted images, to provide anatomical reference for offline motion correction. The interleaved measurement of 71 slices with 10% overlap covered most of the brain in the anterior–posterior direction. Four averages were acquired, resulting in a total acquisition time of 48 min for 71 slices.
software package (Brain Research Institute, Melbourne, Australia, www.brain.org.au/software/). In a separate analysis, multiple orientations of crossing fibers were modeled using constrained harmonized spherical deconvolution (SCD) (41,42) and probabilistic streamline tractography methods were applied. With ultra-high resolution data more fiber orientation distribution (FOD) parameters have to be estimated than actually measured. For this reason, spherical harmonized deconvolution can resolve orientations with smaller angles, and provide a robust estimation of the distribution of fiber orientations in each voxel while preserving angular resolution.
RESULTS Data Processing
In Vivo Imaging at 7T
The DW images were noise cleaned using a two-stage hybrid image restoration procedure (37). In each series, the seven images without diffusion weighting were used to estimate motion correction parameters using rigid-body transformations (38) as implemented in FSL (FMRIB Software Library, University of Oxford 2006, www.fmrib.ox.ac.uk/fsl). The motion correction for the 60 diffusion-weighted images was combined with global registration to the T1 anatomy computed by the same method. The gradient direction for each volume was corrected using the rotation parameters. The registered images were interpolated to the new reference frame, a quantitative T1-weighted anatomical scan with an isotropic voxel resolution of 0.7 mm.
In each voxel, multiple fibers orientations were computed using the ball-and-2-sticks model (39,40). Whole-brain tracking was performed using the MRtrix
Figure 4 shows, in axial sections through the habenula, computed parameter maps of the relaxation times T1, T2*, and proton density (PD). Directly below the ventricular surface, this nucleus can be clearly visualized on both T1 and T2* maps, medial to the caudal part of the dorsal thalamus. The prominent structure shows signal characteristics of both gray matter and white matter. T1 images show a strong contrast of the human habenula with surrounding brain tissue that would typically result from a high myelin content. Probably due to a concomitant high iron content, the habenula also shows reduced T2* compared with neighboring regions. In addition, proton density data from measurements of the regional absolute water content in vivo reveal high contrast between the habenula and adjacent brain. Overall, we can distinguish the human habenula in vivo from surrounding brain tissue based on three different MR contrasts. Table 1 shows the values in human habenula on quantitative parameter maps.
Table 1 Values of the Human Habenula on Quantitative Parameter Maps
Ex Vivo Imaging at 7T
Fiber Tracking
T1 (in ms)
T2* (in ms)
1312–1500 (max. 1900)
28–38
PD (a.u.) 1035–1120 (max. 1335)
Figure 5 illustrates the human habenula ex vivo in the axial plane on maps of T1, T2*, and proton density (PD). Histological studies define the lateral habenula
1022
Strotmann et al.
Figure 5. Habenula ex vivo (axial view): T1, T2*, and proton density maps show distinct lateral and medial habenular nuclei according to myelin stain sections (adapted from 43). MHB, medial habenula; LHB, lateral habenula; hbc, habenular commissure; pc, posterior commissure.
(LHB) and its commissure (hbc) by higher myelin content (included in Fig. 5, of Riley [43]). On axial MR slices, subdivisions of the human habenula become visible. In both T1 and T2* maps the medial and lateral nuclei of the habenula can be distinguished. The Lateral Habenula (LHB) and the habenular commissure (HBC) show a high contrast from surrounding brain and clearly differ from medial habenula (MHB). Overall, in these ex vivo images at 7T, the anatomy of the human habenula can be clearly depicted, and its major subdivisions distinguished. Diffusion Imaging at 7T A probabilistic multi-fiber fit to the 60 direction data set supports more than one fiber orientation in many voxels (Fig. 6). Note the color-coded main fiber orientations with significant signal contribution (threshold: f-value > 0.05) in both the axial plane and, more notably, the coronal plane. Brain sections containing the habenula were superimposed on the quantitative high-resolution T1 map
and compared with histology of stained sections (43,44) and Mai Atlas (27). T1 images show strong contrast between the lateral habenula and the habenular commissure with the surrounding brain tissue, whereas medial habenula shows less contrast. However, both medial and lateral habenula can be clearly seen in the tractography maps. We can distinguish medial habenular nuclei (MBH), having only one fiber orientation in the anterior–posterior direction (green), while lateral habenula (LHB) shows two orientations, anterior–posterior and also superior–inferior (blue), as revealed on the horizontal view. The habenular commissure (HBC) displays a right–left orientation (red), connecting both hemispheres. On coronal view a high level of anatomic detail with corresponding anatomical regions of Mai Atlas (Fig. 7) reveals subdivisions of lateral habenula: fiber populations in LHB do not contribute similarly to the MR signal. Combining T1 contrast and diffusion properties we observe a border region between LHB and MHB: there is a lateral part of LHB with a high myelin content (l-LHB) and a medial part with different diffusion
MRI and DWI of the Human Habenula
1023
Figure 6. Coronal (top) and horizontal (bottom) view of the habenula diffusion with the colorcoded vector orientations showing lateral habenula (LHB), medial (MHB), and the habenular commissure (HBC), and the way of stria medullaris (STM).
properties close to MHB marked with * containing less myelin. Probabilistic tractography shows the two major fiber bundles that connect habenular nuclei with limbic forebrain structures and brainstem on a horizontal slice (Fig. 8). Fibers from LHB appear to pass mainly to the forebrain, while tracts from MHB go down to the brainstem and to its commissure. This is illustrated by streamline tractography using spherical deconvolution. While fibers from LHB seem to run mainly to the forebrain, tracts from MHB go down to the brainstem and to its commissure as illustrated by streamline tractography using spherical deconvolution (Fig. 9). Descending fibers can be detected running from the stria medullaris (STM) to forebrain regions and projections caudally by the way of the fasciculus retroflexus (FR) to paramedian midbrain regions.
DISCUSSION MR imaging at 7T provides sufficient spatial resolution to reveal anatomical details of the human habenula. At ultra-high field strength, there is enough SNR to allow 0.4 mm isotropic voxels in vivo (45–47) for anatomical MRI and MR-Microscopy of ex vivo
brain samples. Using the high field strength of 7T and a combination of zoomed imaging with parallel imaging (ZOOPPA) enables DWI acquisitions with 800 mm to 1 mm isotropic resolution (19). Our data show a high SNR, allowing excellent brain tissue differentiation. Using high performance gradients at 7T, we can resolve fiber crossings at submillimeter resolution and visually identify the human habenula and its substructures in vivo. With spherical harmonic deconvolution and robust probabilistic tractography, complex fiber architecture is revealed in this small region. The internal anatomy of this brain organelle has not previously been investigated using MR techniques. Most of the evidence for the structure and functions of the habenula has been drawn from rodent research, using mainly histological techniques. For the first time, structural MRI reveals the major basic units of the habenular complex: the lateral habenula, the medial habenula, and the habenular commissure. Now, in vivo DWI data discriminate the lateral habenula, which can be clearly distinguished from adjacent brain tissue, and show the main fiber bundles going through this node. The human habenula shows good contrast on both T1 and T2* parameter maps. This suggests strongly that it is characterized by a high concentration both
1024
Strotmann et al.
Figure 7. Coronal view of the habenula diffusion with the color coded vector-orientation zoomed in compared with Mai Atlas (26) showing correspondence. LHb, lateral habenula; MHb, medial habenula; Lim, nucleus limitans; PTc: pretectal area; pc, posterior commissure; Aq, aqueduct; PAG, periaqueductal gray; MD, medial dorsal thalamic nucleus; iml, internal medullary lamina of thalamus, and * a region making border between LHb and MHb.
of myelin (T1 decrease) and of iron (T2* decrease). Ex vivo, the lateral human habenula shows particularly high contrast with surrounding brain tissue on maps of T1 and T2* contrast, as well as on spin density maps. This enables parcellation of both medial and lateral habenular nuclei. Whereas in healthy brain tissue T1 contrast largely depends on myelin, there are several mechanisms
Figure 8. Identification of fiber tracts leaving the habenula. STM, stria medullaris; FR, fasciculus retroflexus.
that can affect T2*. Recent research suggests that increased tissue iron content, in the form of ferretin and/or neuromelanin, is often the reason for a decrease in T2* (48,49). Furthermore, using data from the MP2RAGE sequence, which compensates for B1 field inhomogeneities and tissue heterogeneity, we were able to analyze proton density contrast. Enhanced proton density in the human habenula suggests increased water content. In conclusion, 7T MRI provides good contrast and clear identification of the habenula in vivo, and shows striking internal details in cadaver human brain samples. It is unlikely that similar results could be achieved at lower field strengths. At 7T, we can distinguish lateral, medial habenula, and the habenular commissure. The combined approach of zoomed imaging and parallel imaging (ZOOPPA) was used to acquire diffusion weighted MRI (DWI), which provided exceptionally high resolution, and greatly reduced problems of susceptibility artifacts and image blurring. DWI at 7T can show mainly descending fibers from the habenula, the stria medullaris to forebrain regions and caudally directed projections, in the form of the fasciculus retroflexus to paramedian midbrain
MRI and DWI of the Human Habenula
Figure 9. Corresponding fiber tractography with seed points in lateral habenula (LHB), medial habenula (MHB), and the habenular commissure (HBC).
regions. Overall, we can distinguish lateral and medial habenula, and its commissure, based on high resolution DWI data in vivo. Previously had been mainly reported projections of LHB from basal ganglia to dopaminergic cell regions and limbic afferents of MHB and efferents to serotonergic cell regions and interpeduncular nucleus. LHB showed projections to forebrain regions, whereas MHB showed mainly descending fibers and projections to the habenular commissure. Histology of rat and cat brain shows a heterogeneous lateral habenula (1,50,51): the lateral part (l-LHB) is assumed to be linked to the motor system and the medial part to the limbic system (m-LHB). Future research is planned to investigate structural connectivity and functional activation at high spatial resolution in the human habenula, which may assist in the investigation of the pathophysiology of a wide range of neurologic and psychiatric disorders. In conclusion, further studies of the habenular subdivisions and their role in brain function is likely to improve its understanding and might become crucial for the placement of electrodes for deep brain stimulation in major depression (10). Studies of habenula contrast, as it reflects iron and myelin content, and of its subdivisions, are the subject of further research, which might assist in the understanding of health and clinical disorders. ACKNOWLEDGEMENTS We thank Domenica Wilfing and Elisabeth Wladimirow for MRI scanning and subject recruitment, and Erik T€ urke and Enrico Reimer for technical support. REFERENCES 1. Herkenham M, Nauta WJ. Afferent connections of the habenular nuclei in the rat. A horseradish peroxidase study, with a note on the fiber-of-passage problem. J Comp Neurol 1977;173:123–146. 2. Herkenham M, Nauta WJ. Efferent connections of the habenular nuclei in the rat. J Comp Neurol 1979;187:19–47. 3. Sutherland RJ. The dorsal diencephalic conduction system: a review of the anatomy and functions of the habenular complex. Neurosci biobehav Rev 1982;6:1–13. 4. Hikosaka O, Sesack SR, Lecourtier L, et al. Habenula: crossroad between the basal ganglia and the limbic system. J Neurosci 2008;28:11825–11829.
1025 5. Balcita-Pedicino J, Sesack SR. Ultrastructural analysis of the projection from the lateral hypothalamic area to the lateral habenula in the rat. Soc Neurosci Abstr 2010;36:511–530. 6. Matsumoto M, Hikosaka O. Lateral habenula as a source of negative reward signals in dopamine neurons. Nature 2007;447: 1111–1115. 7. Hikosaka O. The habenula: from stress evasion to value-based decision-making. Nat Rev Neurosci 2010;11:503–513. 8. Morris JS, Smith KA, Cowen PJ, et al. Covariation of activity in habenula and dorsal raphe nuclei following tryptophan depletion. Neuroimage 1999;10:163–172. 9. Sartorius A, Henn FA. Deep brain stimulation of the lateral habenula in treatment resistant major depression. Med Hypotheses 2007;69:1305–1308. 10. Sartorius A, Kiening KL, Kirsch P, et al. Remission of major depression under deep brain stimulation of the lateral habenula in a therapy-refractory patient. Biol Psychiatry 2010;67:e9–e11. 11. Ullsperger M, von Cramon DY. Error monitoring using external feedback: specific roles of the habenular complex, the reward system, and the cingulate motor area revealed by functional magnetic resonance imaging. J Neurosci 2003;23:4308–4314. 12. Salas R, Baldwin P, de Biasi M, et al. BOLD responses to negative reward prediction errors in human habenula. Front Hum Neurosci 2010;4:36. 13. Ide JS, Li CS. Error-related functional connectivity of the habenula in humans. Front Hum Neurosci 2011;5:25. 14. Shelton L, Pendse G, Maleki N, et al. Mapping pain activation and connectivity of the human habenula. J Neurophysiol 2012; 107:2633–2648. 15. Lawson RP, Drevets WC, Roiser JP. Defining the habenula in human neuroimaging studies. Neuroimage 2013:722–727. 16. Savitz JB, Nugent AC, Bogers W, et al. Habenula volume in bipolar disorder and major depressive disorder: a high-resolution magnetic resonance imaging study. Biol Psychiatry 2011;69:336–343. 17. Turner R, Le Bihan D. Single-shot diffusion imaging at 2.0 Tesla. J Magn Reson 1990;86:445–452. 18. Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson 1994;103:247–254. 19. Heidemann RM, Anwander A, Feiweier T, Knosche TR, Turner R. k-space and q-space: combining ultra-high spatial and angular resolution in diffusion imaging using ZOOPPA at 7 T. Neuroimage 2012;60:967–978. 20. Solano-Castiella E, Anwander A, Lohmann G, et al. Diffusion tensor imaging segments the human amygdala in vivo. Neuroimage 2010;49:4:2958–2965. 21. Chanraud S, Reynaud M, Wessa M, et al. Diffusion tensor tractography in mesencephalic bundles: relation to mental flexibility in detoxified alcohol-dependent subjects. Neuropsychopharmacology 2009;34:5:1223–1232. 22. Yeo SS, Chang MC, Kwon YH, et al. Corticoreticular pathway in the human brain: diffusion tensor tractography study. Neurosci Lett 2012;508:9–12. 23. Yang HS, Kwon HG, Hong JH, et al. The rubrospinal tract in the human brain: diffusion tensor imaging study. Neurosci Lett 2011;504:45–48. 24. Marques JP, Kober T, Krueger G, et al. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. Neuroimage 2010;49:1271–1281. 25. Neeb H, Zilles K, Shah NJ. A new method for fast quantitative mapping of absolute water content in vivo. Neuroimage 2006;31: 1156–1168. 26. Warntjes JB, Dahlqvist O, Lundberg P. Novel method for rapid, simultaneous T1, T*2, and proton density quantification. Magn Reson Med 2007;57:528–537. 27. Mai JK, Voss T, Paxinos G, Voss. Atlas of the human brain, 3rd edition. Boston: Elsevier; 2008. p 180–182. 28. Nieuwenhuys R, Voogd J, Van Huijzen C. The human central nervous system. Heidelberg: Springer; 2008. p 79–81. 29. Morelli JN, Runge VM, Feiweier T, et al. Evaluation of a modified Stejskal-Tanner diffusion encoding scheme, permitting a marked reduction in TE, in diffusion-weighted imaging of stroke patients at 3 T. Invest Radiol 2011;45:29–35. 30. Griswold MA, Jakob PM, Chen Q, et al. Resolution enhancement in single-shot imaging using simultaneous acquisition of spatial harmonics (SMASH). Magn Reson Med 1999;41:1236–1245.
1026 31. Feinberg DJ, Hoenninger J, Crooks L, et al. Inner volume MR imaging: technical concepts and their application. Radiology 1985;156:743–747. 32. Mansfield P, Ordidge RJ, Coxon R. Zonally magnified EPI in real time by NMR. J Phys E 1988:275–280. 33. Heidemann RM, Ivanov D, Trampel R, et al. Isotropic sub-millimeter fMRI in the human brain at 7 Tesla: Combining Reduced Field-of-View. Magn Reson Med 2012; 68(5): 1506–1516. 34. Nagy Z, Weiskopf N. Efficient fat suppression by slice-selection gradient reversal in twice-refocused diffusion encoding. Magn Reson Med 2008;60:1256–1260. 35. Ivanov D, Schaefer A, Streicher MN, et al. A simple low-SAR technique for chemical-shift selection with high-field spin-echo imaging. Magn Reson Med 2010;64:319–326. 36. Eichner C, Ivanov D, Feiweier T, et al. High resolution diffusionweighted imaging at 7T - issues of slice thickness and fat suppression. Poster presented at 17th Annual Meeting of the Organization for Human Brain Mapping, Quebec City, QC, Canada, 2011. (abstract 560). 37. Lohmann G, Bohn S, Muller K, et al. Image restoration and spatial resolution in 7-tesla magnetic resonance imaging. Magn Reson Med 2010;64:15–22. 38. Jenkinson M, Bannister P, Brady M, et al. Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 2002;17:825–841. 39. Behrens TE, Berg HJ, Jbabdi S, et al. Probabilistic diffusion tractography with multiple fibre orientations: what can we gain? Neuroimage 2007;34:144–155. 40. Behrens TE, Woolrich MW, Jenkinson M, et al. Characterization and propagation of uncertainty in diffusion-weighted MR imaging. Magn Reson Med 2003;50:1077–1088.
Strotmann et al. 41. Tournier JD, Calamante F, Gadian DG, et al. Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution. Neuroimage 2004;23: 1176–1185. 42. Tournier JD, Calamante F, Connelly A. Robust determination of the fibre orientation distribution in diffusion MRI: non-negativity constrained super-resolved spherical deconvolution. Neuroimage 2007;35:1459–1472. 43. Riley H. (1943). An atlas of the basal ganglia, brain stem and spinal cord. Baltimore, MD: Williams & Wilkins; p 427. 44. Ranft K, Dobrowolny H, Krell D, et al. Evidence for structural abnormalities of the human habenular complex in affective disorders but not in schizophrenia. Psychol Med 2010;40:557–567. 45. Collins CM, Smith MB. Calculations of B(1) distribution, SNR, and SAR for a surface coil adjacent to an anatomically-accurate human body model. Magn Reson Med 2001;45:692–699. 46. Ladd ME. High-field-strength magnetic resonance: potential and limits. Top Magn Reson Imaging 2007;18:139–152. 47. Heidemann RM, Seiberlich N, Griswold MA, et al. Perspectives and limitations of parallel MR imaging at high field strengths. Neuroimaging Clin N Am 2006;16:311–320, xi. 48. Haacke EM, Cheng NY, House MJ, et al. Imaging iron stores in the brain using magnetic resonance imaging. Magn Reson Imaging 2005;23:1–25. 49. Gelman N, Gorell JM, Barker PB, et al. MR imaging of human brain at 3.0 T: preliminary report of transverse relaxation rates and relation to estimated iron content. Radiology 1999;210: 759–767. 50. Andres KH, von D€ uring M, Veh RW. Subnuclear organization of the rat habenular complexes. J Comp Neurol 1999;407:130–150. 51. Iwahori N. A Golgi study on the habenular nucleus of the cat. J Comp Neurol 1977;72:3:319–344.
