Striatal and thalamic automatic segmentation, morphology, and clinical correlates in Parkinsonism: Parkinson ’ s disease, multiple system atrophy and progressive supranuclear palsy

.


Introduction
Parkinson's Disease (PD) is the second most common neurodegenerative disease in the world (De Lau and Breteler, 2006).Alongside PD, there are a range of Parkinsonian disorders, known as Parkinson's plus or atypical parkinsonism, the most common of which are multisystem atrophy (MSA) and progressive supranuclear palsy (PSP).Each is considered a progressively degenerative neurological movement disorder, characterised by variable, but similar, presentations of movement impairment and accompanying cognitive and behavioural neuropsychiatric symptoms such as sleep disturbances, depression, and autonomic dysfunction (Chaudhuri et al., 2006).MSA is subtyped into MSA-cerebellar type and MSA-Parkinsonian type, where MSA-Parkinsonian type is most significantly clinically differentiated from Parkinson's disease through its widespread impacts on autonomic functions (Wenning et al., 2000).PSP can be clinically differentiated from PD through its supranuclear gaze palsy and postural instability (Liscic et al., 2013) which can also occasionally present in MSA.However, these diseases do not always classically present with their characteristic symptom, and instead have significant overlap in their clinical features of bradykinesia, rigidity, tremors, rendering them difficult to discriminate and diagnose in-vivo, with early research showing that 24% of patients were clinically misdiagnosed (Hughes et al., 1992).However, the increasing use of magnetic resonance imaging (MRI) in understanding these conditions has shown that nigrostriatal pathways and cortico-striatal circuits may indeed demonstrate in-vivo differences, beyond the well-established neuropathological changes found on autopsy as alpha-synucleinopathies (Looi et al., 2014;Obeso et al., 2000;Spillantini and Goedert, 2006).
The putamen, caudate nucleus, and thalamus, which all make up aspects of the basal ganglia and the nigrostriatal and cortico-striatopallido-thalamic-cortical circuitry, have varying and important roles in the pathophysiology and development of Parkinsonian diseases (Looi et al., 2014;Obeso et al., 2000).Broadly, the putamen is involved in the development, planning, and implementation of movement (Alexander et al., 1986).The caudate nucleus is associated with emotional regulation, the processing of reward, executive function and decision making (Alexander et al., 1986).The thalamus has a predominantly integratory role in a variety of tasks, such as processing of perception, mediation of motivation and emotions, and in the planning, cognition, and implementation goal directed behaviour (Haber and Calzavara, 2009).The bilateral caudate nucleus and bilateral putamen, which make up the dorsal striatum, will henceforth be referred to as the striatum, for the purposes of this paper.Previous research has shown that decreased stimulation of the fronto-striato-pallido-thalamic and dentatorubrothalamic circuitry is related to atrophy of the thalamus (Looi et al., 2011;Power and Looi, 2015) and other cortical regions (Whitwell et al., 2011) within these circuits.The thalamus and striatum have been previously identified as having potential as biomarkers for neurodegenerative disorders (Looi and Walterfang, 2013;Power and Looi, 2015), especially as related to disease onset (Halliday, 2009;Power and Looi, 2015), progression (Lee et al., 2011;Looi and Walterfang, 2013), and severity (Bohnen and Albin, 2011), and therefore may serve as potent and reliable disease-specific biomarkers for differentiating atypical parkinsonism (Heim et al., 2017;Poewe, 2008).
Regional subcortical and cortical atrophy rates may be more effective at differentiating typical and atypical Parkinsonian disorders than whole-brain atrophy, especially in distinguishing between PSP and MSA (Paviour et al., 2006(Paviour et al., , 2007)).There is research specifically demonstrating differences in volumetric atrophy of the striatum and thalamus in Parkinsonian disorders (Cordato et al., 2005;Ghaemi, 2002;Messina et al., 2011;Nicoletti, 2006;Power and Looi, 2015;Prange et al., 2019;Scherfler et al., 2016;Surova et al., 2015;Worker et al., 2014), whilst other research found no differences (Gellersen et al., 2017;McKeown et al., 2008;Schulz et al., 1999).Furthermore, differential striatal and thalamic morphology between Parkinsonian disorders and controls has been shown (Garg et al., 2015;Hess et al., 2014;McKeown et al., 2008;Nemmi et al., 2015;Sterling et al., 2013;Stezin et al., 2017), though some argue that major morphological differences between patients and controls is not demonstrable (Ghaemi, 2002).Finally, the relationship of structural changes to clinical symptoms yields contrasting findings, with some research finding significant relationships between clinical presentation and greater atrophy or morphological change of the striatum and thalamus (Chung et al., 2017;Geevarghese et al., 2015;Sterling et al., 2013), and other research yields no such results (Garg et al., 2015).There are complex structural-functional interactions between the striatum, the thalamus, and various other brain regions that may lead to differing outcomes.This study seeks to better understand the relationship between clinical illness severity measures, striatal and thalamic brain region volumetry, and morphology.Neurodegeneration in vivo, quantified by striatal and thalamic volume and shape, may be key to distinguishing these disorders, and perhaps predicting neuropsychiatric performance, and eventually providing definitive diagnosis.
The aim of this current paper is to investigate the volume and shape of the striatum and thalamus in PD, MSA, and PSP patients, compared to healthy controls.We investigated these differences through automatic segmentation and shape analysis of these regions of interest from participant structural MRI data.We hypothesised that there would be significant differences in striatal and thalamic volume and shape between the patient groups.The neurodegenerative disease profiles of these disorders led to the hypothesis that the striatal and thalamic volumes and shape would be most robust in healthy controls, followed by Parkinson's disease, then MSA patients, and finally PSP patients with the most regional atrophy and shape deflation.We sought to determine if the differences in volumetry are associated with any clinical changes in behaviour, cognition, or motor function, and whether they could be used as predictive measures of neuropsychiatric symptoms.We hypothesised that reduction of volume in all regions of interest, regardless of diagnosis, would be correlated with worse outcome measures on disease severity scales, cognitive function scales, and for neuropsychiatric symptom scales.

Participants
This study included 67 participants 45 -85 years of age (32 male and 33 female), 18 individuals diagnosed with PD, 12 diagnosed with MSA, 16 diagnosed with PSP, and 21 aged-matched healthy controls, recruited by the Hospital Clinico San Carlos (HCSC) and Hospital Severo Ochoa in Madrid, Spain.One PSP patient was excluded due to a changed diagnosis to corticobasal degeneration during a follow up assessment, and one healthy control excluded due to unreadable MRI data (incorrect transformation of image that could not be corrected).Due to the small sample size, the MSA patients, who were originally classified into subgroups (cerebellar MSA and Parkinsonian MSA), were collated into a single diagnostic group for statistical analysis.PSP patients were not clinically subtyped during data collection, and so remain as a single diagnostic group for statistical analyses.The clinical diagnoses were made by a specialist neurologist at HCSC, with PD diagnosed with London Brain Bank criteria, PSP with National Institute of Neurological Disorders and Stroke and Society criteria, and MSA with the Gilman criteria (Gilman et al., 2008;Hughes et al., 1992;Litvan et al., 1996).Ethics approval was obtained by Comité Ético de Investigación Cliníco Hospital and the Australian National University Human Research Ethics Committee.All participants gave informed written consent prior to participating in the study.

Clinical assessment of patients
Clinical measures of parkinsonism, its severity, and cognitive and behavioural (neuropsychiatric) function were assessed in patients, using Spanish translations of the Hoehn-Yahr Scale (Goetz et al., 2004), Unified Parkinson's Disease Rating Scale (UPDRS) (Goetz et al., 2008), Mini Mental-State Examination (MMSE) (Folstein et al., 1975), Neuropsychiatric Inventory (NPI) (Cummings, 2020), and Frontal Assessment Battery (FAB) (Dubois et al., 2000).The patients' number of years since diagnosis was reported, as was age and sex for all participants.Healthy controls were not assessed for clinical measures of Parkinsonism.

MRI acquisition, processing, and volumetric/shape analysis
T1-weighted MRI data were acquired for patients and healthy controls by a 3.0T Phillips MR scanner.FSL_anat tool was used to process T1 images and were reoriented to MNI T1 1 mm standard space (using fslreorient2std) (Jenkinson et al., 2012;Smith et al., 2004).The images were then cropped to reduce field-of-view-size, and excessive non-brain tissue removed.FSL-FAST was used to perform a bias-field correction (RF/B1-inhomogeneity-correction), before FSL-FLIRT and FSL-FNIRT was performed for registration to transform images to standard space using linear and nonlinear registration tools, respectively.Images were then processed again to remove all non-brain tissue using FSL-BET (Smith, 2002).Tissue-type segmentation was then performed to produce masks for grey matter, white matter, and cerebrospinal fluid (CSF) using FSL-FAST, followed by subcortical automatic segmentation using FSL-FIRST (Smith, 2002).Mean voxel partial volume estimates were extracted using FSL-Stats (in mm3) for grey matter, white matter, and CSF, and each was multiplied by the volume of tissue type, to quantify the intracranial volume (ICV) (Buckner et al., 2004).The regions of interest (ROI) for this paper's analyses included the bilateral caudate nucleus, bilateral thalamus, and bilateral putamen.
The automatic bilateral subcortical structure segmentations of the relevant ROIs, created using FSL-FIRST software, were used in statistical analyses, after being multiplied by the scaling factor.These segmentations were also included in a shape analysis using the SPHARM-PDM module (spherical harmonic point distribution method -UNC, Chapel Hill) in 3D Slicer (ww.slicer.org)via the SlicerSALT project (Vicory et al., 2018) to generate ROI shapes.The first step fills any interior holes of the binary segmentations to ensure spherical topology, before surface mesh and parametrisation spheres are generated.These outputs are then used to compute corresponding spherical harmonic descriptions before being sampled into triangulated surfaces which can be used in subsequent analyses.Quality control of the shapes was manually investigated using ShapePopulationViewer.The average shape of each subcortical ROI from the control group was computed using the Shape Variation Analyzer module of the SlicerSALT project, to be included as the comparison shape in subsequent statistical analyses.

Statistical analyses
Statistical analyses were performed using SlicerSALT and IBM SPSS Statistics for Macintosh version 27 (IBM Corp., Armonk, N.Y., USA).Analysis of variance (ANOVA) was conducted to determine any interaction between sex or age and diagnosis between groups.Multivariate analysis of covariance (MANCOVA) was conducted to determine significant differences between participant groups (PD, MSA, PSP, and healthy controls) in the volumes of bilateral caudate nucleus, putamen, and thalamic structures.The covariates included in analysis were age, sex, and total ICV to account for head size.Planned simple contrast analyses, using Bonferroni correction, were run to examine the comparison between patient groups and controls.For shape analyses, we used the Covariate Significance Testing module of SlicerSALT, with a family-wise error (FWE) correction for multiple comparisons, with age and sex included as covariates, to compare patient groups and controls.Hierarchical regression analyses were conducted to assess whether volume significantly differentiated patient groups above and beyond covariates, and whether clinical measures added any value to differentiating groups.A Pearson partial correlation was conducted to determine whether volume correlated with the outlined clinical measures, and whether any correlation remained whilst accounting/controlling for the effect of patient diagnoses.

Participant demographics
Demographics from participants are listed in Table 1.There were no significant differences in age (F = 0.59, p = 0.623) or sex (χ 2 = 4.03, p = 0.259) between patient groups and controls.There were significant differences (p < 0.001) between patient groups in the Hoehn-Yahr Scale, Universal Parkinson's Disease Rating scale, and the Frontal Assessment Battery measures, with PSP patients having the worst scores on all three measures and PD patients having the best.

Volumetric analyses
The MANCOVA determined that there was a significant difference in all the ROIs between groups (p = 0.045, F = 0.169), using Pillai's Trace statistic, when accounting for age, sex, and total ICV.The results of the MANCOVA between subjects effect comparing the groups for each region is displayed in Table 2.The planned Bonferroni corrected pairwise comparisons, found that PSP ROI volumes, when compared with healthy control ROI volumes, were significantly different.Specifically, the PSP patients had significant reduced volume when compared to the control group for the left thalamus (p = 0.017, mean difference: − 702.87 mm 3 ), the right thalamus (p = 0.011, mean difference: − 742.06 mm 3 ), and the right putamen (p = 0.032, mean difference: − 518.34 mm 3 ).There were no other between group comparisons that were statistically significant for ROI volumetric analyses.

Shape analyses
The bilateral thalamus, bilateral caudate, and left putamen had statistically significant (FWE p < 0.05) morphological shape differences between patient groups and controls.Figures of each of these shape differences are displayed below (Figs.1-3).
The antero-ventral medial portion of the left and right caudate head varied significantly (L: p = 0.004, R: p = 0.012) between groups (Fig. 1).This effect was driven by differences between PSP patients and controls (p = 0.004) and from MSA patients (p = 0.014), whereas PD patients did not significantly differ compared to controls.For the left putamen, the ventro-posterior aspect differed significantly (p = 0.028), (shown in Fig. 2) and was driven by PSP (p < 0.001) and PD (p < 0.001) compared to controls, but not MSA patients (p > 0.05).The left antero-ventral thalamus (p = 0.004), and the right dorsolateral thalamus (p = 0.012) were significantly different between groups (Fig. 3), which was bilaterally driven by differences in PSP patients compared to controls (L: p = 0.002, R: p < 0.001), rather than PD (L: p = 0.006, R: p > 0.05) and MSA (L: p > 0.05, R: p = 0.016) patients.

Clinical correlation analyses
Regression analyses determined that age, sex, and clinical measures did not predict diagnosis (p = 0.103), and that ROI volume did not predict diagnosis above and beyond the clinical scores (p = 0.143).Pearson correlation analyses showed the left thalamus (p = 0.044), bilateral caudate (L: p = 0.017, R: p = 0.027), and bilateral putamen (L: p = 0.009, R: p = 0.027) significantly correlated with total neuropsychiatric inventory (NPI) scores, of which only the left caudate and left putamen remained significant (p = 0.025 and p = 0.025 respectively), when controlling for diagnosis.

Volumetric interpretation
Overall, the volume of the bilateral thalamus and right putamen were significantly different between patients and controls, with patient groups having significantly reduced volume of each ROI.Planned comparisons determined that PSP patients consistently had the most atrophy across ROIs compared to controls.Our finding is consistent with literature that notes that PSP patients tend to have more significant neurodegeneration compared to PD, with more significant atrophy in the putamen, thalamus, caudate nucleus, and globus pallidus found in neuroimaging (Lee et al., 2013).There was no significant reduction in caudate volume across patient groups when compared to controls.Although the planned comparisons found no significance between groups, we do note that PSP patients non-significantly trend towards having more caudate nucleus atrophy than other patient groups and controls.This non-significant result is consistent with previous research finding caudate atrophy does not always differentiate patient groups (Cordato et al., 2002).However, other studies demonstrated differences in caudate volume for all patient groups relative to controls (Owens-- Walton et al., 2018;Schulz et al., 1999), with larger sample sizes than the present study, which may be why we were unable to reproduce these findings.Furthermore, automatic segmentation of the caudate has been  shown to generate smaller volumes and a lower percentage volume difference compared to manual delineation (Mansoor et al., 2021;Perlaki et al., 2017), which may be why our results follow the established trends but are not statistically significant.We found a significant reduction in volume of the right putamen for PSP patients compared to controls.The left putamen volume followed the same pattern, with PSP patients having the most putaminal atrophy relative to other patient groups and controls, but this was not statistically significant.Previous research has robustly demonstrated putaminal atrophy in PSP patients (Focke et al., 2011;Looi et al., 2011;Messina et al., 2011), especially relative to controls.However, putaminal atrophy does not always differentiate Parkinsonian disorders (Massey et al., 2012;Messina et al., 2011).Due to the significant involvement of dopaminergic neurons within the putamen, these results are consistent with the current understanding of significant degeneration of dopamine transporters in PSP found post-mortem (Ruberg et al., 1985).The inability to demonstrate the bilateral atrophy of the putamen in our PSP patient group relative to controls and other patient groups may be due to our limited sample size.Additionally, our PD and MSA patients had comparable results to the control group, with PD patients non-significantly showing less atrophy bilaterally compared to MSA patients.Research has demonstrated MSA patients tend to have atrophy in the putamen, more so than PD patients (Huppertz et al., 2016;Sako et al., 2014), which is in contradiction to our current findings, however this may be due to patients being in the later stages of their disease, smaller sample sizes, and the combined grouping of the disparate MSA subtypes within our study.
Thalamic atrophy has been well demonstrated in PSP patients when compared to controls, and this has been replicated in our study (Boxer et al., 2006;Cordato et al., 2005;Messina et al., 2011;Schofield et al., 2011).Notably, Messina et al. did find that most atrophy occurred in PSP patients, then MSA patients, followed by PD and controls, but our results found more significant atrophy in PD patients relative to MSA patients.This may again be due to our PD patients being in later disease stages, MSA subtypes being combined, and our more limited sample size (Messina et al., 2011).Prior research has suggested that thalamic atrophy occurs due to decreased stimulation to the thalamus from greater neural circuitry dysfunction (Looi et al., 2011;Power and Looi, 2015) and therefore may underpin the atrophy demonstrated in our study.

Shape analysis interpretation
There were significant overall group differences in the bilateral head of caudate nucleus between patient groups and controls, with the left caudate demonstrating more significant differences between the groups.PSP patients had the most significant differences in the bilateral caudate compared to controls, PD and MSA patients.The anterior portion of the caudate nucleus is functionally useful in cognitive and emotional processing of reward stimuli due to its relationship with lateral and medial prefrontal cortices (Driscoll et al., 2021).These findings are consistent with other research which notes greater loss of dopamine transporters in the anterior caudate for PSP patients (Oh et al., 2012(Oh et al., , 2011)).The implication of changes in these regions of the head of the caudate in our PSP patients relative to controls demonstrates the possible loss of connections from frontal cortices resulting in the clinical features displayed.These results validate other research in the field which showed deflation of caudate morphometry most significantly for PSP patients (Looi et al., 2011).
The shape of the ventro-posterior aspect of left putamen was significantly different between patient groups and controls, with the PSP patients most significantly differing from controls.The pathophysiology behind this may be related to the demonstrated loss of dopamine transporters in Parkinsonian disorders, where the ventral putamen showed significantly more loss in PSP and MSA patients, with relative sparing in PD patients (Oh et al., 2012).Additionally, putaminal shape change relative to controls may arise from cortico-striatal and fronto-striatal circuit dysfunction, which are linked to impacts on motor function and frontal eye fields in PSP (Alexander et al., 1986;Utter, 2008).Furthermore, a dorsal-ventral pattern of atrophy in the striatum has been described in the disease process of PSP (Looi and Walterfang, 2013), and is consistent with the findings in our study.
The antero-ventral left thalamus and the dorsolateral right thalamus significantly varied between patient groups and controls, with PSP patients most significantly different from controls, relative to MSA and PD.These replicate findings from manual tracing of thalamic volumes and subsequent shape analysis demonstrating significant antero-ventral and ventrolateral thalamic deflation for PSP patients (Power et al., 2017).The ventrolateral thalamus receives significant output from the basal ganglia, and is involved in projection to motor and premotor cortices (Haber and Calzavara, 2009).The relationships and function of the dorsolateral nucleus of the thalamus are yet to be fully understood, however it is considered to be involved with the limbic system and spatial learning and memory (Bezdudnaya and Keller, 2008), all of which are known to be impaired in Parkinsonian presentations.Thalamic volumetric atrophy and the related morphological changes seen in our PSP patients is consistent with previous findings that these changes may be due to degradation of circuitry, like the dentato-rubro-thalamic tract, which are connected to these thalamic nuclei (Hess et al., 2014;Surova et al., 2015).

Clinical correlation interpretation
We investigated whether clinical features of Parkinsonian disorders were related to volumetric structural changes, as the cognitive profile of the disorders alone may not provide enough clinical utility in differentiating them (Lee et al., 2012).Our study was not able to demonstrate that volumetric atrophy of the striatum or thalamus could predict diagnosis beyond clinical measures, age, sex, and ICV, which has been found in other research (Schulz et al., 1999).However, we were able to determine a correlational relationship between the volume of the bilateral striatum and left thalamus with total NPI scores.The left caudate and left putamen volumes were also significantly associated with NPI scores when controlling for diagnosis.The NPI is an assessment of functional behavioural disturbances, including delusions, anxiety, disinhibition, and aberrant motor activity (Cummings et al., 1994).Atrophy of the striatum and thalamus would broadly impact upon a multitude of circuits which cause cognitive, emotional, and motor dysfunction assessed using the NPI.Our findings support the research that demonstrates a relationship between striatal atrophy and clinical measures, however, there are variable findings on the relationship between the striatal volumetry and functional clinical scales.Some studies demonstrated a relationship between striatal atrophy in PSP (Josephs et al., 2011) and PD (Owens-Walton et al., 2018) with neuropsychiatric measures of Frontal Behavioural Inventory, and UPDRS and MMSE, respectively.In contrast to these findings, other research noted that PSP cortical atrophy did not correlate with any cardinal clinical features (Schofield et al., 2011), that cognition generally showed extra-striatal degeneration rather than degeneration in the putamen (Broussolle et al., 1999), and that putaminal or caudal volume and measures of general cognitive or executive function had no relationship in PD (Lewis et al., 2016;Mak et al., 2015;Nemmi et al., 2015).

Limitations and future research
Our major limitations are due to the small sample size within our analyses, stemming from the relatively low prevalence of PSP and MSA.However, our sample size is comparable to other research in this domain (Hess et al., 2014;Looi et al., 2011;Messina et al., 2011;Nemmi et al., 2015;Whitwell et al., 2011), which characterises the difficulty in recruiting patients for large scale neuroimaging studies in rarer diseases.Notably, the sample sizes for PSP and MSA were smaller than for PD patients and controls, but remain comparable to other studies that include these patient groups as part of their analyses.Additionally, further investigation of the magnitude of shape differences (deflation or inflation) could not be conducted due to data analysis limitations that arose due to the COVID-19 public health restrictions and will be clarified in subsequent analyses.
This study was able to investigate clinical correlates with automatic segmentation of volumes associated with Parkinsonian disorders, however further research could be done to investigate correlates with shape analyses.Additional research investigating connectivity through functional neuroimaging between and within the identified regions, and the clinical correlates of that functional connectivity could elucidate a clearer picture of the implicated structures in Parkinsonian disorders.Research using different neuroimaging modalities (such as diffusion tensor imaging) could additionally yield more information on the interaction of neural regions and symptoms to differentiate the disorders.Finally, pending the sharing of compatible data, uniform imageprocessing, and analysis methodology, neuroimaging meta-analysis such as those conducted by the ENIGMA consortium might be possible for PSP and other atypical Parkinsonian disorders.

Conclusion
In conclusion, we have found significant volumetric atrophy and morphometric changes of the striatum and thalamus in PSP patients in contrast to healthy age-matched controls, but which did not significantly differentiate this patient group from PD and MSA patients.This paper concords with previous research that there is neurodegeneration outside of the basal ganglia in Parkinsonian disorders, as well as significant atrophy and shape change in striatum and thalamus, which may relate to neuropsychiatric symptoms.There is promising evidence for these neural regions as biomarkers for PSP in the future.Future prospective research should focus on differentiating PSP from the almost clinically indistinguishable PD, and MSA, through a combination of larger sample sizes, structural and functional neuroimaging, and their clinical correlations.This will aid better diagnosis and prognosis for patients suffering these neurodegenerative diseases.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Shape analysis of patient groups & controls of bilateral caudate.The colour variation on the bilateral caudate shown denotes areas that are significantly different between patient groups relative to controls, with greater differences noted from yellow to red (p < 0.05).The caudate is displayed from a variety of angles (L caudate shown from anterior and inferior aspects, R caudate shown from posterior and lateral aspects).L: Left, R: right.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Fig. 2 .
Fig. 2. Shape analysis of patient groups & controls of left putamen.The colour variation on the left putamen shown denotes areas that are significant different between groups, with greater differences noted from yellow to red (p < 0.05).The putamen is displayed from anterior and inferior aspects.L: Left.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Fig. 3 .
Fig. 3. Shape analysis of patient groups & controls of bilateral thalamus.The colour variation on the bilateral thalamus shown denotes areas that are significantly different between groups, with greater differences noted from yellow to red (p < 0.05).The thalamus is displayed from a variety of angles (L thalamus shown from anterior and inferior aspects, R thalamus shown from inferior and lateral aspects).L: Left, R: right.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Table 2
Multivariate between group volumetric differences.Volume in mm 3 is listed for each region of interest per patient group.L: Left, R: Right PD: Parkinson's disease, PSP: Progressive supranuclear palsy, MSA: Multiple system atrophy.* denotes significant differences between noted groups in additional planned Bonferroni contrasts.