A probabilistic atlas of the cerebellar white matter

YNIMG-12571; No. of pages: 9; 4C: 2, 3, 5, 7, 8 NeuroImage xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

1

Full Length Article

2Q1

A probabilistic atlas of the cerebellar white matter

F O

5 6 7 8 9

a

1 0

a r t i c l e

11 12 13 14

Article history: Received 21 June 2015 Accepted 7 September 2015 Available online xxxx

15 16 17 18 19 20 21

Keywords: Cerebellum White matter Atlas Parcellation map Probabilistic tractography Human Connectome Project

Department of Neurosurgery, Radboud University Medical Centre, Huispostnummer 636, Geert Grooteplein-Zuid 10, 6525 GA Nijmegen, The Netherlands Oxford Centre for Functional Magnetic Resonance Imaging of the Brain, John Radcliffe Hospital, Oxford OX3 9DU, United Kingdom Department of Anatomy, Radboud University Medical Centre, Geert Grooteplein-Zuid 10, 6525 GA Nijmegen, The Netherlands d Donders Institute for Brain, Cognition and Behavior, Kapittelweg 29, 6525 EN Nijmegen, The Netherlands b

i n f o

R O

c

a b s t r a c t

D

P

Imaging of the cerebellar cortex, deep cerebellar nuclei and their connectivity are gaining attraction, due to the important role the cerebellum plays in cognition and motor control. Atlases of the cerebellar cortex and nuclei are used to locate regions of interest in clinical and neuroscience studies. However, the white matter that connects these relay stations is of at least similar functional importance. Damage to these cerebellar white matter tracts may lead to serious language, cognitive and emotional disturbances, although the pathophysiological mechanism behind it is still debated. Differences in white matter integrity between patients and controls might shed light on structure–function correlations. A probabilistic parcellation atlas of the cerebellar white matter would help these studies by facilitating automatic segmentation of the cerebellar peduncles, the localization of lesions and the comparison of white matter integrity between patients and controls. In this work a digital three-dimensional probabilistic atlas of the cerebellar white matter is presented, based on high quality 3 T, 1.25 mm resolution diffusion MRI data from 90 subjects participating in the Human Connectome Project. The white matter tracts were estimated using probabilistic tractography. Results over 90 subjects were symmetrical and trajectories of superior, middle and inferior cerebellar peduncles resembled the anatomy as known from anatomical studies. This atlas will contribute to a better understanding of cerebellar white matter architecture. It may eventually aid in defining structure–function correlations in patients with cerebellar disorders. © 2015 Published by Elsevier Inc.

E

4

K.M. van Baarsen a,⁎, M. Kleinnijenhuis b, S. Jbabdi b, S.N. Sotiropoulos b, J.A. Grotenhuis a, A.M. van Cappellen van Walsum c,d

42 40 39

R

41

Introduction

44

Atlases of the cerebellar cortex and cerebellar nuclei are of great value, but they are not sufficient to map cerebellar anatomy. The white matter that connects these relay stations in the cerebrocerebellar circuitry is also vital to function. In order to determine white matter structure–function relationships, dedicated cerebellar white matter atlases are needed.

49

U

47 48

N C O

43

45 46

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38

R

E

C

T

3Q2

Abbreviations: DTI, diffusion tensor imaging; DWI, diffusion weighted imaging; FA, fractional anisotropy; FMRIB, Oxford Centre for Functional MRI of the Brain; FNIRT, FSL non-linear registration tool; FSL, FMRIB Software Library; HCP, Human Connectome Project; ICP, inferior cerebellar peduncle; IIT, Illinois Institute of Technology; JHU, Johns Hopkins University; MCP, middle cerebellar peduncle; MNI, Montreal Neurological Institute; MRI, magnetic resonance imaging; ROI, region of interest; SCP, superior cerebellar peduncle; SEM, standard error of the mean; SUIT, spatially unbiased infratentorial template. ⁎ Corresponding author. E-mail addresses: [email protected] (K.M. van Baarsen), [email protected] (S. Jbabdi), [email protected] (S.N. Sotiropoulos), [email protected] (J.A. Grotenhuis), [email protected] (A.M. van Cappellen van Walsum).

Structural and functional imaging studies of the cerebellar cortex and deep cerebellar nuclei, as well as cerebellar connectivity studies are gaining attraction due to the important but previously underestimated role of the cerebellum in non-motor function (De Smet et al., 2013; Marien et al., 2014; Stoodley, 2012; Stoodley and Schmahmann, 2009; Strick et al., 2009). In fact, damage to the cerebellar white matter has been associated with many cognitive, language and emotional disorders, i.e. in spinocerebellar degeneration, autism spectrum disorder and cerebellar mutism (Becker and Stoodley, 2013; Kawai et al., 2009; Kuper and Timmann, 2013). However, the exact pathogenesis is still under discussion. On the one hand, cerebellar cortical areas seem to be directly involved in language and cognitive functions (Stoodley and Schmahmann, 2009). White matter lesions may disconnect these areas. On the other hand, cerebellar white matter lesions might lead to a hypoactivity in supratentorial areas of the brain by disruption of the cerebello-cerebral circuitry, resulting in language and cognitive dysfunction (Catsman-Berrevoets and Aarsen, 2010; van Baarsen and Grotenhuis, 2014). The relationship between type and location of the white matter lesion, supra- or infratentorial cortical hypoactivity and clinical deficits has not been clarified yet.

http://dx.doi.org/10.1016/j.neuroimage.2015.09.014 1053-8119/© 2015 Published by Elsevier Inc.

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

R O

O

F

2

81 82 83 84

T

C

E

R

80

R

78 79

O

76 77

MRI studies (Diedrichsen et al., 2009). In 2011, they added a probabilistic atlas of the cerebellar nuclei (Diedrichsen et al., 2011). Although the above mentioned atlases facilitate segmentation of cerebellar lobules and nuclei, they do not cover the cerebellar white matter. Ideally, researchers would provide detailed information on the exact lobule or peduncle measurements are taken in. However, locating each cerebellar area in every subject is very time-consuming and requires a thorough understanding of cerebellar anatomy. A digital atlas would allow the automatic localization of white matter structures in the cerebellum, resulting in easier and more efficient segmentations. Further, the atlas would provide a template facilitating betweengroup analyses in clinical studies. Data from large cohorts of patients with cerebellar disorders and healthy subjects are needed in order to compare the anatomy and integrity of white matter fibers in both

C

74 75

To allow optimal analysis and interpretation of cerebellar studies, atlases of the cerebellar cortex and nuclei are used to locate regions of interest. Duvernoy's Atlas provides beautiful anatomical images of brain stem and cerebellum, but the information is two-dimensional and not in standard space (Duvernoy et al., 2009). Schmahmann et al. (1999) provided the first digital three-dimensional atlas of cerebellar lobules and fissures in the stereotaxic space of Talairach and Tournoux (Schmahmann et al., 1999). In 2006, Diedrichsen et al. developed a Spatially Unbiased Infratentorial Template for the cerebellum (SUIT), to overcome problems with anatomical inaccuracy in the cerebellar region with the commonly used MNI template (Diedrichsen, 2006). Based on the 1999 atlas of Schmahmann et al, Diedrichsen et al. also developed a cerebellar cortical parcellation atlas in standard space (MNI as well as their own SUIT template) which is convenient in cerebellar functional

N

72 73

U

71

E

D

P

Fig. 1. Regions of interest used for tractography, overlaid on the 1 mm T1-weighted MNI template. A. Regions of interest for estimating the superior cerebellar peduncle (SCP). Top panel: thalamic subnuclei from the Oxford Thalamic Connectivity Atlas (Behrens et al., 2003), pictured in dark red, were used as waypoint and stop masks. Bottom: right and left dentate nuclei depicted in khaki green were used as seed masks. B. Regions of interest for estimating the middle cerebellar peduncle (MCP). The belly of the pons was segmented in the sagittal plane at X = 88 for the right MCP (yellow) and X = 91 for the left MCP (orange), which were used as seed masks. The Probabilistic Atlas of the Cerebellum (Diedrichsen et al., 2011), thresholded at 50%, was chosen as a waypoint and stop mask (blue). An exclusion mask was drawn in the mid-sagittal plane at X = 89 (red) to prevent streamlines from crossing to the contralateral hemisphere. C. Regions of interest for the inferior cerebellar peduncle (ICP). Seed regions in the restiform body (bright green) were placed at Z = 24 based on the 1 mm FA skeleton. Again, the Probabilistic Atlas of the Cerebellum (Diedrichsen et al., 2011), thresholded at 50%, was chosen as waypoint and stop region (blue) and the mid-sagittal exclusion mask was used. A second exclusion mask was located in the tonsils to prevent streamlines from crossing the cerebello-medullary fissure (pink).

Fig. 2. Correspondence of tractography results with textbook anatomy. A and C: final probabilistic atlases showing the three cerebellar peduncles. A, view from left posterolateral side matches the left-lateral view from panel B, representing the anatomy of the three cerebellar peduncles as it appears in textbooks. With permission from (Nieuwenhuys et al., 2008). C, left cerebellar peduncles from a posteromedial view. Note how the inferior cerebellar peduncle is squeezed between superior and middle ones. Orange: superior cerebellar peduncle; green: middle cerebellar peduncle and blue: inferior cerebellar peduncle. Arrowhead points at the aberrant tract of the left superior cerebellar peduncle. In A, MCP and ICP are made translucent for better visualization, but no thresholds were used.

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

85 86 87 88 89 90 91 92 93 94 95 96 97 98

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

3

R O

O

F

et al., 1994). Fiber tracking algorithms use information provided by these directional diffusion measurements to reconstruct trajectories of fiber tracts (Mori et al., 1999). At present, tractography has allowed the visualization of major fiber tracts in the human brain (Catani and Thiebaut de Schotten, 2008; Mori et al., 2008; Oishi et al., 2008; Zhang et al., 2011). Only a few tractography based white matter atlases actually include the cerebellum. Most of them are unable to demonstrate the decussation of the superior cerebellar peduncles (SCP) at the level of the mesencephalon (Mori et al., 2008; Peng et al., 2009; Zhang et al., 2011). As is known from dissection and histological studies, left and right SCP decussate in the mesencephalon and traverse toward the contralateral thalamus (Dejerine and Dejerine-Klumpke, 1901; Nieuwenhuys et al., 2008; Stilling, 1878) (Mollink et al., 2015). The absence of the decussation is probably due to the use of the diffusion tensor model for fiber orientation estimation in these studies (Wakana et al., 2004). This method is known for its inability to accurately represent crossing fiber regions. Furthermore, a deterministic approach to tractography is often used, which does not explicitly account for measurement inaccuracies and uncertainties (Behrens et al., 2007). In 2014, Varentsova et al. developed a probabilistic version of the Illinois Institute of Technology (IIT) atlas (Varentsova et al., 2014). Data were derived from multiple subjects, however these diffusion weighted data were of relatively low angular resolution. To overcome this limitation, data from multiple subjects were combined, leading to a dataset with diffusion weighted measurements along a large number of different directions. Fiber orientation distributions were estimated using constrained spherical deconvolution (Tournier et al., 2004, 2007). This method extracts more directional information from the diffusion signal than the diffusion tensor model and allows the estimation of fiber crossing patterns. Indeed, the IIT atlas is the first to demonstrate fiber crossings at the decussation of the SCPs. Unfortunately, that information is not included in their parcellation map (https:// www.nitrc.org/projects/iit2). It follows that current atlases have certain limitations. First, they are developed by averaging multiple diffusion datasets before performing

D

E

T

C

108

E

106 107

R

104 105

R

102 103

N C O

100 101

groups. However, to be able to combine and compare data from multiple subjects, measurements should be made in a common reference space. Ideally, the same template should be used by all investigators to facilitate comparisons not only between groups of subjects, but also between studies. At present, the only method to non-invasively investigate the threedimensional architecture of white matter tracts and their integrity invivo, is Diffusion Weighted Imaging (DWI). This MRI technique is based on the diffusivity of water molecules, being larger in the direction parallel to the axons compared to the perpendicular direction (Basser

U

99

P

Fig. 3. Tractography results of left inferior (ICP) and middle cerebellar peduncle (MCP) in 3D. Left posterolateral view of the inferior (blue) and middle (green) cerebellar peduncle based on the group probability maps of 90 subjects. No thresholds were applied. Note that the ICP fibers run primarily toward the anterior cerebellar lobules (black arrowhead), whereas MCP fibers run toward posterior lobules (white arrowhead).

Fig. 4. Aberrant tracts of the superior (A–C) and inferior (D–F) cerebellar peduncles. From left to right transverse, coronal and sagittal images of the final probabilistic atlas are shown. A, B and C: bilaterally, aberrant tracts (white arrowheads) run posteromedial to the true superior cerebellar peduncles (black arrowheads). Bilateral aberrant tracts seem to connect at the level of the posterior commissure. The (group-) probability in the majority of voxels is below 0.15. Crosshair coordinates are X = 94, Y = 99 and Z = 69. D, E and F: aberrant tracts run bilaterally from the inferior cerebellar peduncle directly toward the flocculus, traversing the cerebellomedullary fissure. Probability values are up to 0.3 which is higher than in the aberrant fibers of the SCP. Crosshair coordinates are X = 105, Y = 86 and Z = 27. For these images, no thresholds were set.

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144

Subjects & MRI data

176 177

189

MRI datasets of 90 healthy volunteers in the age range of 22 to 35 (mean age 30), were obtained from the Human Connectome Project database (http://www.humanconnectome.org) (Van Essen et al., 2013). The sample included 55 females and 35 males. Seventy-eight subjects were right-handed, eight were left-handed and three subjects were ambidextrous. None of the subjects had prior diagnosis of neurological or psychiatric disease and no structural abnormalities were found in their scans. Subjects were scanned on a customized Siemens 3 T “Connectome Skyra” scanner at Washington University at a notably high resolution. Main diffusion parameters include a multiband factor of 3 (Moeller et al., 2010), nominal voxel size of 1.25 × 1.25 × 1.25 mm, and 270 diffusion weighted scans (each acquired twice) equally distributed over 3 shells defined with b-values of 1000, 2000 and 3000 s/mm2 (Sotiropoulos et al., 2013; Van Essen et al., 2013).

190

Data processing

191

The diffusion data were preprocessed using the HCP diffusion pipelines (Glasser et al., 2013; Sotiropoulos et al., 2013). Briefly, distortions were corrected using a model-based approach that simultaneously considers susceptibility and eddy-current induced distortions, as well as head motion (Andersson and Sotiropoulos, 2015; Andersson and Sotiropoulos, 2015a, 2015b). Fiber orientations were estimated using a parametric deconvolution approach that accounts for crossing fibers and partial volume, as well as the non-monoexponential decay in b-space (Jbabdi et al., 2012). Bayesian inference of the model parameters and automatic relevance determination of the crossing patterns was performed using the relevant FSL toolbox (Behrens et al., 2007). The estimated orientations were then used for reconstructing cerebellar white matter tracts using probabilistic streamline tractography.

170 171

178 179 180 181 182 183 184 185 186 187 188

192 193 194 195 196 197 198 199 200 201 202 203

T

168 169

C

166 167

E

164 165

R

162 163

R

160 161

O

158 159

C

156 157

N

154 155

U

152 153

F

175

151

O

Methods

149 150

R O

174

147 148

E

172 173

tractography (Mori et al., 2008; Peng et al., 2009; Varentsova et al., 2014; Zhang et al., 2011). This may lead to a loss of information regarding the intersubject anatomical variability. Preferably, tractography should be performed individually in every subject followed by averaging of these tractography results. Second, previous atlases are based on the Montreal Neurological Institute (MNI) template. This template is mainly developed for supratentorial gray matter, with little anatomical detail in the cerebellum. The loss of detailed information is probably due to individual differences in the angle between the brainstem and cerebellum (Diedrichsen, 2006). Third, most previous atlases have not been able to demonstrate the correct anatomical position of the SCPs (Catani and Thiebaut de Schotten, 2008; Mori et al., 2008; Salamon et al., 2007; Wakana et al., 2004). Fourth, most atlases use deterministic tractography and/or the diffusion tensor models, thus not accounting for uncertainty at the individual level and potential complex fiber patterns. The one study that does use probabilistic tractography is based on tractography in a single high-angular resolution dataset in standard space that was derived from combining multiple subject low-angular data, but its parcellation map does not incorporate the crossing of the SCPs (Varentsova et al., 2014). We present an atlas that overcomes these shortcomings. It contains a digital three-dimensional parcellation map of the cerebellar white matter, based on high quality (high spatial and angular resolution) 3 T diffusion MRI data from 90 subjects participating in the Human Connectome Project (Sotiropoulos et al., 2013; Van Essen et al., 2013). The probabilistic information is represented in the most commonly used MNI template system, but to overcome the earlier mentioned limitations associated with MNI it is also presented in the spatially unbiased infratentorial (SUIT) template (Diedrichsen, 2006).

P

145 146

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

D

4

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

5

R O

O

F

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

205

Regions of interest (ROI), which were used for tractography, are shown in Fig. 1. They were segmented either using existing atlases or drawn manually on structural images or fractional anisotropy (FA) maps from the 1 mm non-linear MNI template using ITK-SNAP (Yushkevich et al., 2006). T1 and DWI volumes of individual subjects were registered to the 1 mm MNI template using the HCP pipelines (Glasser et al., 2013). For the superior cerebellar peduncle (SCP), the seed region was chosen in the ipsilateral dentate nucleus (Fig. 1A, bottom). The contralateral thalamus (Fig. 1A, top), taken from the Oxford Thalamic Connectivity Atlas (Behrens et al., 2003) at a 50% threshold, was chosen as waypoint as well as stop region (i.e. streamlines had to go through the thalamus to be considered valid and they were terminated upon entering the thalamus). For the middle cerebellar peduncle (MCP), the seed region was chosen in the ipsilateral pons (at X = 88 for right and X = 91 for left MCP) (Fig. 1B, top and bottom). The non-linear 1 mm Probabilistic Atlas of the Cerebellum (Diedrichsen et al., 2009), representing the average cerebellar gray matter probability, thresholded at 50%, was chosen as waypoint as well as stop region (Fig. 1B, bottom). An exclusion mask in the midline of the posterior fossa was used to prevent streamlines from crossing to the contralateral side, either through the pons or through the cerebellar cortex which is continuous through both hemispheres (Fig. 1B, middle). For the inferior cerebellar peduncle (ICP), the seed region was chosen in the ipsilateral restiform body (Fig. 1C, top, green ROI). Again, the non-linear 1 mm Probabilistic Atlas of the Cerebellum (Diedrichsen et al., 2009), thresholded at 50%, was chosen as waypoint as well as stop region and the posterior fossa midline was chosen as exclusion mask. Finally, to prevent streamlines from entering the cerebellum directly from the restiform body by crossing the cerebellomedullary fissure, an exclusion mask covering the tonsils, which were extracted from the Probabilistic Atlas of the Cerebellum (Diedrichsen et al., 2009), was used (Fig. 1C). Tractography was performed in individual subjects using the probabilistic implementation in FSL (Behrens et al., 2007). Default settings were used: curvature threshold 78°; 5000 streamlines per

216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241

C

E

214 215

R

212 213

R

210 211

N C O

208 209

U

206 207

voxel; maximal number of steps 2000; step length 0.5 mm; “loopcheck” to terminate streamlines that loop back on themselves and “one way condition” to ensure that the exclusion/inclusion criteria are applied to the streamlines regardless of their starting orientation polarity. For the SCP, a maximum of 275 steps (13.5 cm path length) was used to prevent streamlines from passing between thalamic subnuclei via the tectal plate to the contralateral side. The “one way condition” was left out for tractography of the ICPs in order to demonstrate streamlines at both sides of the seed ROI in the brain stem. Tractography was performed for every peduncle (left and right SCP, left and right MCP and left and right ICP) separately, resulting in 6 different output volumes per subject. The path distribution output for each peduncle was represented in the MNI space (same space as the input ROIs). For each voxel, the ratio of the number of traversing streamlines relative to the total number of generated valid streamlines was computed, resulting in “path probability maps” for each peduncle.

242

Atlas creation

258

For every subject, the path probability maps of superior, middle and inferior cerebellar peduncles were binarized, based on a threshold at their 50th percentile. Averaging the binary maps of all 90 subjects resulted in the final probabilistic atlas, in which each voxel represents the proportion of subjects in whom that voxel actually belongs to a specific tract (i.e., the probability, expressed as a ratio between 0 and 1). This probability is different from the path probability, which is the streamline ratio as determined by tractography. The final probabilistic atlas included all 90 subjects. To examine the convergence of tractography results with an increasing sample size, the averaging of the path probability maps was evaluated in bins of ten subjects. For each peduncle, average path probability maps were calculated from 10, 20, …, 90 subjects. The absolute differences between subsequent group-averaged path probability maps were calculated per voxel. The percentage change, expressed as the sum over all voxels in the path difference map divided by the sum over all voxels in the previous path probability map, is a measure of the effect of adding another ten subjects in the development of the final atlas. The threshold chosen to indicate convergence was 5%. To

259

E

Tractography

T

204

D

P

Fig. 5. Illustration of the digital three-dimensional parcellation map of the cerebellar white matter, derived by applying thresholds of 10, 50 and 90% of its robust range to the final (group-) probabilistic atlas. Orange: superior cerebellar peduncle; green: middle cerebellar peduncle; blue: inferior cerebellar peduncle. Coordinates are in MNI space, background is the 1 mm T1 weighted MNI template. A, transverse sections from cranial to caudal. The decussation of the superior cerebellar peduncles is visible at Z = 55. B, coronal sections from anterior to posterior. The decussation of the superior cerebellar peduncles is visible at Y = 100. C, sagittal sections from right to left. Note the left–right symmetry.

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

243 244 245 246 247 248 249 250 251 252 253 254 255 256 257

260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277

6

278 279

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

287

Results

288 289

325 326

The cerebellar peduncles were successfully reconstructed in all subjects. Tractography results strikingly matched the layered white matter anatomy as known from classical dissection studies (Fig. 2). The SCP (or brachium conjunctivum) laid deepest, being situated in the upper roof of the fourth ventricle. The SCPs crossed to the contralateral thalamus at the level of the mesencephalon. The course of the SCPs at the level of the decussation was close to horizontal. The ICP (or corpus restiforme), which has a cylindrical aspect at the level of the medulla, flattens at the level of the inferior part of the fourth ventricle and is squeezed between the superior and middle cerebellar peduncles before its fibers fan out toward the anterior vermis and lobules (Fig. 3). More ICP fibers entered the anterior lobules (lobule I–V, anterior to the primary fissure) (black arrowheads), compared to the posterior lobules (lobule VI to VIII; posterior to the primary fissure) (white arrowheads). On the other hand, the majority of fibers from the MCP (or brachium pontis) run to the posterior rather than anterior lobules (Fig. 3). This was assessed by calculating the mean path probability per voxel in anterior versus posterior lobules (lobule I–V and lobule VI–VIII, respectively) in the average path probability map over all 90 subjects. For the right and left MCPs, the mean path probability was 0.0057 (SD 0.0103) and 0.0082 (SD 0.0142) in anterior lobules, whereas in posterior lobules it was almost three times as high: 0.0198 (SD 0.0257) and 0.020 (SD 0.0289). For the right and left ICPs, on the other hand, the mean path probability was 0.0035 (SD 0.0138) and 0.0084 (SD 0.0219) in anterior lobules, whereas in posterior lobules it was more than three times as low: 0.0009 (SD 0.0018) and 0.0013 (SD 0.0022). The final probabilistic atlas showed two aberrant fiber tracts at low probability thresholds (Fig. 4). Bilaterally, streamlines were generated from the ICP seed region, crossing the lateral cerebellomedullary fissure into the flocculus. Further, aberrant streamlines branched out from the dorsal aspect of the SCPs traveling upward at the posterolateral side of the aqueduct toward the thalamus (Figs. 2 and 4). The final probabilistic atlas was thresholded at 10, 50 and 90% of its robust range to make the binary parcellation maps (Fig. 5). Regarding the individual tractography results, a considerable variability was observed in the total number of streamlines between subjects as well as between tracts, demonstrating the need for inclusion of a large number of subjects. In line with the size of the seed and target

t1:1 t1:2 t1:3 t1:4

Table 1 Streamline total counts and ratios for the cerebellar peduncles. First column: mean number of retained streamlines in 90 subjects and corresponding standard deviations (second column). Third and fourth column show the mean path probability (number of streamlines per voxel, relative to the total number of retained streamlines in the tract) and its standard deviation, when path probability maps are averaged over all subjects.

305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324

t1:5

Total number of retained streamlines (N = 90)

t1:6 t1:7 t1:8 t1:9 t1:10 t1:11 t1:12

O

R O

P

D

303 304

E

301 302

T

300

C

298 299

E

296 297

R

294 295

R

292 293

O

290 291

C

284

N

282 283

U

280 281

ROIs, the right and left MCP had the largest number of retained streamlines per subject. The amount of streamlines that was generated in superior and inferior cerebellar peduncles was comparable (Table 1). In the MCP and ICP, the number of streamlines was normally distributed, whereas in the SCP the frequency distribution was skewed to the left, indicating an increasing number of subjects with a decreasing number of streamlines. However, after normalization of the number of streamlines per voxel to path probabilties, and averaging these path probabilities of all 90 subjects, the SCP appeared to have the highest path probability per voxel when averaged over all non-zero voxels (Table 1). As expected, increasing the number of subjects contributing to the atlas resulted in convergence of the data (Fig. 6). After 80 subjects, the effect of another batch of 10 subjects was 5% or less for all tracts. The change in voxel-wise standard error of the mean (SEM) is shown in Fig. 7, demonstrating stabilization over the extent of the peduncles after ~80 subjects. In this study, a new white matter atlas for the human cerebellum was proposed, based on high quality diffusion data of 90 subjects from the Human Connectome Project. Trajectories of superior, middle and inferior cerebellar peduncles strongly resemble anatomy as known from textbooks and are highly symmetrical (Figs. 2 and 5). For the ICP, path probability values were higher in the anterior lobules compared to the posterior lobules. The reverse was the case for the MCP (Fig. 3). This is in correspondence with the current hypotheses on topographical organization in the cerebellum, such that the anterior lobe and lobule VIII contain the representation of the “sensorimotor cerebellum” and lobules VI and VII of the posterior lobe comprise the “cognitive cerebellum” (Stoodley and Schmahmann, 2010). Anterior lobules receive sensorimotor input from the body through the ICPs, and posterior lobules receive input from motor and non-motor areas of the cerebral cortex through corticopontocerebellar fibers (Dum and Strick, 2003). At low probability thresholds, aberrant tracts had been generated from the ICP, crossing the lateral cerebellomedullary fissure into the flocculus, as well as from the dorsal side of the SCP into the mesencephalic tectum (Fig. 4). This is probably due to limitations in resolution and ability to resolve crossing fibers. Partial volume in voxels in the cerebellomedullary fissure may cause streamlines to “jump over” from brain stem to flocculus. Likewise, streamlines from the SCP may be erroneously tracked into the adjacent central tegmental tract or medial longitudinal fasciculus. However, the majority of these aberrant tracts do only appear at low probabilities. The MCP, which is the largest of the peduncles, had the most streamlines per voxel. However, when corrected for the total number of retained streamlines per tract, it was the SCP that had the highest path probability per voxel (Table 1). This is probably caused by the convergence of streamlines from the dentate nucleus through the narrow upper roof of the fourth ventricle into the mesencephalon. The low mean probability per voxel in the MCP, on the other hand, may be due to the fact that after leaving the seed region in the pons, the streamlines

F

285 286

confirm this convergence, the voxel-wise standard error of the mean was calculated for subsequent maps. For visualization purposes and practical application, the groupprobabilistic atlases were binarized into parcellation maps based on thresholds of 10, 50 and 90% of their “robust intensity range”, which calculates values similar to the 2% and 98% percentiles. The MNI compatible parcellation maps were then non-linearly transformed into the SUIT template using the SUIT toolbox version 2.7 (http://www.icn.ucl.ac.uk/ motorcontrol/imaging/suit.html) (Diedrichsen, 2006).

Right SCP Left SCP Right MCP Left MCP Right ICP Left ICP

Path probability (number of streamlines per voxel) relative to (total number of retained streamlines)

Mean

SD

Mean

SD

65.223 58.113 684.607 674.813 41.443 74.844

54.746 51.079 139.206 132.308 13.148 15.607

0.19 0.20 0.02 0.02 0.03 0.03

0.86 0.88 0.10 0.09 0.25 0.21

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

7

R O

O

F

directly toward the cerebellar cortex. As a consequence, the cerebellar tonsils were excluded from the white matter atlas. This shortcoming to the atlas was deemed acceptable, because damage to the cerebellar tonsils does not seem to have immediate neurological consequences. Empirically, it is known that they may be surgically removed without neurological sequelae. Further, a meta-analysis of functional MRI studies could not discover neurological functions that were exclusively ascribed to the cerebellar tonsils (Stoodley and Schmahmann, 2009). It may however reflect their connectivity to the cingulate, precuneus and other areas of the default-mode network (Krienen and Buckner, 2009). Another item that may have affected the results is that ROIs were drawn in MNI standard space and transformed to individual space before performing tractography. With possible variation in registration accuracy, this may have resulted in ROIs that deviate from their intended anatomical structures in individual space, resulting in different numbers of generated streamlines across subjects. This problem was accounted for by normalizing the number of streamlines per voxel to path probabilities. In addition, the effect of normalization and registration errors should have been minimized by the large sample size. Finally, we have used the HCP pipelines for registration that allows very accurate transformations between subject and standard spaces (Glasser et al., 2013). There was an imbalance in male versus female and left- versus righthanded subjects. Especially this last aspect is of interest considering the known lateralization in cerebellar functions (Hu et al., 2008; Marien et al., 2001). One might anticipate differences in the cerebellar paths related to handedness or gender, based on previous studies that have already shown asymmetries, especially in the cerebrum (Amunts et al., 2000; Park et al., 2004). However, the goal of this study was to develop a probabilistic atlas of the structural paths that is applicable to the general population. Future studies will address specifically this limitation and may explore how white matter volume and diffusion measures are correlated to specific cerebellar cortical functions and lateralization. Compared to the currently available cerebellar parcellation maps (Mori et al., 2008; Varentsova et al., 2014), our results show a slightly more lateral position of the SCP in the roof of the fourth ventricle. Further, our parcellations extend further into the cerebellar hemispheres as well as the mesencephalon and clearly demonstrate the decussation of the SCPs (Fig. 8). Perhaps the biggest advantage of our atlas is the fact that it is based on individual probabilistic tractography in a very large population of 90 subjects. Therefore, apart from the parcellation map, this atlas provides information on the probability per voxel that it actually contains fibers of a certain tract. Employing probabilistic tractography and state-ofthe-art data quality allowed us to track reproducibly across subjects, including the decussation of the SCPs. This atlas may be of great value in analyzing white matter integrity changes in cerebellar disorders such as tumor, stroke, or degeneration.

388 389 390 391 392 393 394 395 396 397 398

D

E

T

C

E

386 387

R

384 385

R

382 383

N C O

380 381

diverge widely into the surrounding cerebellar cortex. The convergence of the SCP, as well as the divergence of the MCP and ICP, thus reflects the white matter anatomy as currently known (Mollink et al., 2015; Nieuwenhuys et al., 2008). This atlas was based on diffusion data of 90 subjects, which is the largest cohort ever used for preparation of a white matter atlas. It was also shown that with a cohort of 80 subjects, the effect of adding another ten subjects is less than 5% (Fig. 6). An additional analysis showed an obvious decrease of the Standard Error of the Mean per voxel (Fig. 7). Therefore, the cohort of 90 subjects was deemed large enough for a reliable population-based probability atlas of the cerebellar white matter. The anatomical position of the chosen inclusion and exclusion masks may be one of the most influential parameters in tractography. The high quality of the diffusion data made it possible to reconstruct the tracts without a lot of adjustments in tractography settings. However, the use of a midline exclusion mask through the posterior fossa prohibited the visualization of the anterior and posterior cerebellar commissures (Angevine et al., 1961; Salamon et al., 2007). These fiber bundles are very small and in any case would have been hard to detect with tractography. Another exclusion mask was chosen in the cerebellar tonsils to prevent streamlines from crossing the cerebellomedullary fissure

U

378 379

P

Fig. 6. Percentage difference between subsequent path probability maps (group-averaged over an increasing number of subjects), expressed as the sum of the difference values between subsequent maps, divided by the sum of all values in the previous map. After 80 subjects, the modulating effect of another batch of 10 subjects is 5% or less.

Fig. 7. Heatmaps of the standard error of the mean of subsequent group-averaged path probability maps. The SEM obviously decreases for SCP (left panel), MPC (middle panel) and ICP (right panel) with an increasing number of subjects.

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

E

D

P

R O

O

F

8

447

C

T

Fig. 8. Comparison of the parcellation map (at 10% probability threshold) presented in this study to the ones currently available. A, parcellation map of this study, showing the decussation of the SCPs (top panel; green and purple) and white matter segmentations, in particular the middle cerebellar peduncle, extending further into the cerebellar hemispheres (lower panel; blue) compared to the other maps. B, parcellation map of JHU (Johns Hopkins University) (Mori et al., 2008) in which a differentiation of the crossing pathways within the decussation of the SCPs is lacking (upper panel; pink and orange). C, Parcellation map of IIT (Illinois Institute of Technology) (Varentsova et al., 2014) in which the color coding represents the most probable gray matter connection in that voxel. Note that right and left middle cerebellar peduncles (yellow) have the same color and the same segmentation intensity, suggesting one fiber tract instead of two separate tracts. Also, differentiation of the crossing pathways within the decussation of the SCPs is lacking. Top panel, Z = 54; middle panel, Z = 40; lower panel, Z = 34.

461

Conclusion

462

470

This study presents a white matter atlas of the cerebellum, based on probabilistic tractography of high resolution, high quality diffusion MR data of 90 healthy subjects from the Human Connectome Project database. The atlas comes as a 3D probabilistic map, as well as binary parcellations at different probability thresholds. MNI—as well as SUIT—compatible versions of the atlas are freely available at www. nitrc.org (Neuroimaging Informatics Tools and Resources Clearinghouse) and www.nccn.nl (Neurosurgical Centre Nijmegen, The Netherlands).

471

Conflict of interest

457 458

463 464 465 466 467 468 469

472

R

R

O

455 456

C

453 454

N

451 452

U

450

E

459 460

The atlas will facilitate the segmentation of the cerebellar peduncles, the localization of lesions and the comparison of white matter integrity between patients and controls by simply applying the atlas' ROIs and extracting diffusion measures for each subject. It may eventually aid in defining structure–function relationships in patients with cerebellar disorders. The high quality of these diffusion data evokes for detailed cerebellar studies investigating smaller white matter fiber tracts such as the cerebellar commissures and differentiating between climbing and mossy fibers and the corticonuclear projections. Future work may also investigate connectivity between cerebellar cortex and olivary, pontine, dentate or thalamic nuclei, throwing light on the cerebello-cerebral circuitry and the cerebellar organization of sensorimotor and cognitive functions.

448 449

The authors declare that there is no conflict of interest.

Acknowledgments

473

Data were provided by the Human Connectome Project, WU-Minn Consortium (Principal Investigators: David Van Essen and Kamil Ugurbil; 1U54MH091657) funded by the 16 NIH Institutes and Centers that support the NIH Blueprint for Neuroscience Research; and by the McDonnell Center for Systems Neuroscience at Washington University.

474

Appendix A. Supplementary data

479

475 476 477 478

Supplementary data to this article can be found online at http://dx. 480 doi.org/10.1016/j.neuroimage.2015.09.014. 481 References

482

Amunts, K., Jancke, L., Mohlberg, H., Steinmetz, H., Zilles, K., 2000. Interhemispheric asymmetry of the human motor cortex related to handedness and gender. Neuropsychologia 38, 304–312. Andersson, J.L., Sotiropoulos, S.N., 2015a. Non-Parametric Representation and Prediction of Single- and Multi-Shell Diffusion-Weighted MRI Data Using Gaussian Processes. Neuroimage. Under minor revision. Andersson, J.L.R., Sotiropoulos, S.N., 2015b. An integrated approach to correction for offresonance effects and subject movement in diffusion MR imaging. NeuroImage 30 (Epub ahead of print). Angevine, Mancall, Yakovlev, 1961. The Human Cerebellum. An Atlas of Gross Topography in Serial Sections first edition. Little, Brown and Company, Boston. Basser, P.J., Mattiello, J., LeBihan, D., 1994. MR diffusion tensor spectroscopy and imaging. Biophys. J. 66, 259–267. Becker, E.B., Stoodley, C.J., 2013. Autism spectrum disorder and the cerebellum. Int. Rev. Neurobiol. 113, 1–34. Behrens, T.E., Johansen-Berg, H., Woolrich, M.W., Smith, S.M., Wheeler-Kingshott, C.A., Boulby, P.A., Barker, G.J., Sillery, E.L., Sheehan, K., Ciccarelli, O., Thompson, A.J., Brady, J.M., Matthews, P.M., 2003. Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat. Neurosci. 6, 750–757.

483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

K.M. van Baarsen et al. / NeuroImage xxx (2015) xxx–xxx

F

O

R O

P

D

E

T

C

E

R

R

Mori, S., Oishi, K., Jiang, H., Jiang, L., Li, X., Akhter, K., Hua, K., Faria, A.V., Mahmood, A., Woods, R., Toga, A.W., Pike, G.B., Neto, P.R., Evans, A., Zhang, J., Huang, H., Miller, M.I., van Zijl, P., Mazziotta, J., 2008. Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template. NeuroImage 40, 570–582. Nieuwenhuys, R., Voogd, J., van Huijzen, C., 2008. The Human Central Nervous System. fourth edition. Springer-Verlag, Berlin Heidelberg New York. Oishi, K., Zilles, K., Amunts, K., Faria, A., Jiang, H., Li, X., Akhter, K., Hua, K., Woods, R., Toga, A.W., Pike, G.B., Rosa-Neto, P., Evans, A., Zhang, J., Huang, H., Miller, M.I., van Zijl, P.C., Mazziotta, J., Mori, S., 2008. Human brain white matter atlas: identification and assignment of common anatomical structures in superficial white matter. NeuroImage 43, 447–457. Park, H.J., Westin, C.F., Kubicki, M., Maier, S.E., Niznikiewicz, M., Baer, A., Frumin, M., Kikinis, R., Jolesz, F.A., McCarley, R.W., Shenton, M.E., 2004. White matter hemisphere asymmetries in healthy subjects and in schizophrenia: a diffusion tensor MRI study. NeuroImage 23, 213–223. Peng, H., Orlichenko, A., Dawe, R.J., Agam, G., Zhang, S., Arfanakis, K., 2009. Development of a human brain diffusion tensor template. NeuroImage 46, 967–980. Salamon, N., Sicotte, N., Drain, A., Frew, A., Alger, J.R., Jen, J., Perlman, S., Salamon, G., 2007. White matter fiber tractography and color mapping of the normal human cerebellum with diffusion tensor imaging. J. Neuroradiol. 34, 115–128. Schmahmann, J.D., Doyon, J., McDonald, D., Holmes, C., Lavoie, K., Hurwitz, A.S., Kabani, N., Toga, A., Evans, A., Petrides, M., 1999. Three-dimensional MRI atlas of the human cerebellum in proportional stereotaxic space. NeuroImage 10, 233–260. Sotiropoulos, S.N., Jbabdi, S., Xu, J., Andersson, J.L., Moeller, S., Auerbach, E.J., Glasser, M.F., Hernandez, M., Sapiro, G., Jenkinson, M., Feinberg, D.A., Yacoub, E., Lenglet, C., Van Essen, D.C., Ugurbil, K., Behrens, T.E., Consortium, W.U.-M.H., 2013. Advances in diffusion MRI acquisition and processing in the Human Connectome Project. NeuroImage 80, 125–143. Stilling, B., 1878. Neue Untersuchungen über den Bau des kleinen Gehirns des Menschen. In: Fischer, Cassel (Ed.), Dritter Band. Stoodley, C.J., 2012. The cerebellum and cognition: evidence from functional imaging studies. Cerebellum 11, 352–365. Stoodley, C.J., Schmahmann, J.D., 2009. Functional topography in the human cerebellum: a meta-analysis of neuroimaging studies. NeuroImage 44, 489–501. Stoodley, C.J., Schmahmann, J.D., 2010. Evidence for topographic organization in the cerebellum of motor control versus cognitive and affective processing. Cortex 46, 831–844. Strick, P.L., Dum, R.P., Fiez, J.A., 2009. Cerebellum and nonmotor function. Annu. Rev. Neurosci. 32, 413–434. Tournier, J.D., Calamante, F., Gadian, D.G., Connelly, A., 2004. Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution. NeuroImage 23, 1176–1185. Tournier, J.D., Calamante, F., Connelly, A., 2007. Robust determination of the fibre orientation distribution in diffusion MRI: non-negativity constrained super-resolved spherical deconvolution. NeuroImage 35, 1459–1472. van Baarsen, K.M., Grotenhuis, J.A., 2014. The anatomical substrate of cerebellar mutism. Med. Hypotheses 82, 774–780. Varentsova, A., Zhang, S., Arfanakis, K., 2014. Development of a high angular resolution diffusion imaging human brain template. NeuroImage 91, 177–186. Wakana, S., Jiang, H., Nagae-Poetscher, L.M., van Zijl, P.C., Mori, S., 2004. Fiber tract-based atlas of human white matter anatomy. Radiology 230, 77–87. Yushkevich, P.A., Piven, J., Hazlett, H.C., Smith, R.G., Ho, S., Gee, J.C., Gerig, G., 2006. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. NeuroImage 31, 1116–1128. Zhang, S., Peng, H., Dawe, R.J., Arfanakis, K., 2011. Enhanced ICBM diffusion tensor template of the human brain. NeuroImage 54, 974–984.

N C O

615

Behrens, T.E., Berg, H.J., Jbabdi, S., Rushworth, M.F., Woolrich, M.W., 2007. Probabilistic diffusion tractography with multiple fibre orientations: what can we gain? NeuroImage 34, 144–155. Catani, M., Thiebaut de Schotten, M., 2008. A diffusion tensor imaging tractography atlas for virtual in vivo dissections. Cortex 44, 1105–1132. Catsman-Berrevoets, C.E., Aarsen, F.K., 2010. The spectrum of neurobehavioural deficits in the Posterior Fossa Syndrome in children after cerebellar tumour surgery. Cortex 46, 933–946. De Smet, H.J., Paquier, P., Verhoeven, J., Marien, P., 2013. The cerebellum: its role in language and related cognitive and affective functions. Brain Lang. 127, 334–342. Dejerine, J., Dejerine-Klumpke, A., 1901. Anatomie des Centres Nerveux. J. Rueff, Paris. Diedrichsen, J., 2006. A spatially unbiased atlas template of the human cerebellum. NeuroImage 33, 127–138. Diedrichsen, J., Balsters, J.H., Flavell, J., Cussans, E., Ramnani, N., 2009. A probabilistic MR atlas of the human cerebellum. NeuroImage 46, 39–46. Diedrichsen, J., Maderwald, S., Kuper, M., Thurling, M., Rabe, K., Gizewski, E.R., Ladd, M.E., Timmann, D., 2011. Imaging the deep cerebellar nuclei: a probabilistic atlas and normalization procedure. NeuroImage 54, 1786–1794. Dum, R.P., Strick, P.L., 2003. An unfolded map of the cerebellar dentate nucleus and its projections to the cerebral cortex. J. Neurophysiol. 89, 634–639. Duvernoy, H.M., Delman, B.N., Sorensen, A.G., Kollias, S.S., Haacke, E.M., 2009. Duvernoy's Atlas of the Human Brain Stem and Cerebellum. Springer-Verlag, Vienna. Essen, Van, Smith, D.C., Barch, S.M., Behrens, D.M., Yacoub, T.E., Ugurbil, E., Consortium, W.U.-M.H., 2013. The WU-Minn Human Connectome Project: an overview. NeuroImage 80, 62–79. Glasser, M.F., Sotiropoulos, S.N., Wilson, J.A., Coalson, T.S., Fischl, B., Andersson, J.L., Xu, J., Jbabdi, S., Webster, M., Polimeni, J.R., Van Essen, D.C., Jenkinson, M., Consortium, W.U.-M.H., 2013. The minimal preprocessing pipelines for the Human Connectome Project. NeuroImage 80, 105–124. Hu, D., Shen, H., Zhou, Z., 2008. Functional asymmetry in the cerebellum: a brief review. Cerebellum 7, 304–313. Jbabdi, S., Sotiropoulos, S.N., Savio, A.M., Grana, M., Behrens, T.E., 2012. Model-based analysis of multishell diffusion MR data for tractography: how to get over fitting problems. Magn. Reson. Med. 68, 1846–1855. Kawai, Y., Suenaga, M., Watanabe, H., Sobue, G., 2009. Cognitive impairment in spinocerebellar degeneration. Eur. Neurol. 61, 257–268. Krienen, F.M., Buckner, R.L., 2009. Segregated fronto-cerebellar circuits revealed by intrinsic functional connectivity. Cereb. Cortex 19, 2485–2497. Kuper, M., Timmann, D., 2013. Cerebellar mutism. Brain Lang. 127, 327–333. Marien, P., Engelborghs, S., Fabbro, F., De Deyn, P.P., 2001. The lateralized linguistic cerebellum: a review and a new hypothesis. Brain Lang. 79, 580–600. Marien, P., Ackermann, H., Adamaszek, M., Barwood, C.H., Beaton, A., Desmond, J., De Witte, E., Fawcett, A.J., Hertrich, I., Kuper, M., Leggio, M., Marvel, C., Molinari, M., Murdoch, B.E., Nicolson, R.I., Schmahmann, J.D., Stoodley, C.J., Thurling, M., Timmann, D., Wouters, E., Ziegler, W., 2014. Consensus paper: Language and the cerebellum: an ongoing enigma. Cerebellum 13, 386–410. Moeller, S., Yacoub, E., Olman, C.A., Auerbach, E., Strupp, J., Harel, N., Ugurbil, K., 2010. Multiband multislice GE-EPI at 7 tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magn. Reson. Med. 63, 1144–1153. Mollink, J., van Baarsen, K.M., Dederen, P.J.W.C., Foxley, S., Miller, K.L., Jbabdi, S., Slump, C.H., Grotenhuis, J.A., Kleinnijenhuis, M., van Cappellen, A.M., van Walsum, 2015. Dentatorubrothalamic tract localization with post mortem MR Diffusion Tractography compared to histological 3D reconstruction. Brain Structure & Function. Under Minor Revision. Mori, S., Crain, B.J., Chacko, V.P., van Zijl, P.C., 1999. Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann. Neurol. 45, 265–269.

U

502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558

9

Please cite this article as: van Baarsen, K.M., et al., A probabilistic atlas of the cerebellar white matter, NeuroImage (2015), http://dx.doi.org/ 10.1016/j.neuroimage.2015.09.014

559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614