ParkCore: Harmonized analysis of neuroimaging datasets from NorthAmerican & Indian Parkinson studies
Poster No:
173
Submission Type:
Abstract Submission
Authors:
Nikhil Bhagwat1, Shweta Prasad2, Michelle Wang1, Brent McPherson1, Sebastian Urchs1, Jitender Saini2, Pramod Pal2, Ravi Yadav2, Edward Fon1, Madeleine Sharp1, Alain Dagher1, Jean-Baptiste Poline1
Institutions:
1McGill University, Montreal, Quebec, 2NIMHANS, Bangalore, Karnataka
First Author:
Co-Author(s):
Introduction:
Parkinson's disease (PD) is a neurodegenerative movement disorder with increasing global prevalence and societal burden. The significant heterogeneity in symptom and neurological presentations has hindered the development of reliable diagnostic and progression biomarkers. Further, despite growing availability of large-scale PD studies across the globe, data aggregation and analytic comparisons have been limited due to lack of standardization across acquisition protocols, image processing, and statistical analysis. In the ParkCore project (Fig.1), we compare neuroanatomical phenotypes across multiple PD cohorts from the US, Canada, and India by 1) harmonizing imaging and clinical variables and 2) standardizing MRI processing. We highlight the commonalities and differences in brain morphometry in PD patients resultant of acquisition and biological factors.
Methods:
We harmonize and process age-matched samples from PPMI (n=294) (Marek et al., 2018), QPN (n=162) (Gan-Or et al., 2020), and three NIMHANS (n=132, 91, 295) cohorts (Prasad et al., 2022). Harmonization of demographic and clinical variables is performed using Neurobagel (https://neurobagel.org/) tools. The MRI data are acquired on different scanners but scans are processed identically using Nipoppy workflows. We quantify 1) cortical thickness (CTh) and subcortical volumes using FreeSurfer-7 (Fischl, 2012), 2) cerebellar lobular volumes using the MAGeT Brain pipeline (Pipitone et al., 2014), 3) structural connectomes based on white-matter tracts using TractoFlow (Theaud et al., no date), and 4) functional networks using fMRIPrep (Esteban et al., 2019) and Nilearn pipelines. We calculate regional CTh and subcortical volumes using DKT parcellation (Klein and Tourville, 2012). To control for site effects, we assess CTh and volumetric differences of PD-vs-control separately for the three cohorts using GLM. We control for age and sex in all models and additionally for total-intracranial-volume in regional and total-cerebellar-volume in cerebellar volumetric analyses.
Results:
Fig.2 shows distributions for several image derived phenotypes (IDPs) for the various cohorts. NIMHANS-1 shows scanner related bias with higher CTh values on average. The PD-vs-control comparisons show significant differences in CTh of posterior-cingulate (right) in both QPN and NIMHANS cohorts. Significant differences in Thalamic (bilateral) volume are seen in PPMI, but not in QPN or NIMHANS. PD cohorts consistently show higher CSF volumes but are not statistically significant.
Conclusions:
Data harmonization and standardized processing is essential to minimize methodologically induced phenotypic variations in cross-cohort comparisons. Image acquisition - scanners and protocols - seem to play a stronger role in cortical thickness quantification compared to the regional volumetry. Scanner specific image normalization of MRI data and analysis of clinical phenotypes are needed (ongoing) to address cohort-specific feature shifts and isolate reliable PD-specific neurological signatures across datasets. In the future we will work to make these distributed results searchable with the Neurobagel project.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 1
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping 2
Subcortical Structures
White Matter Anatomy, Fiber Pathways and Connectivity
Keywords:
Degenerative Disease
Movement Disorder
STRUCTURAL MRI
1|2Indicates the priority used for review
Provide references using author date format
Fischl, B. (2012) ‘FreeSurfer’, NeuroImage, 62(2), pp. 774–781.
Gan-Or, Z. et al. (2020) ‘The Quebec Parkinson Network: A Researcher-Patient Matching Platform and Multimodal Biorepository’, Journal of Parkinson’s disease, 10(1), pp. 301–313.
Klein, A. and Tourville, J. (2012) ‘101 labeled brain images and a consistent human cortical labeling protocol’, Frontiers in neuroscience, 6, p. 171.
Marek, K. et al. (2018) ‘The Parkinson’s progression markers initiative (PPMI) - establishing a PD biomarker cohort’, Annals of clinical and translational neurology, 5(12), pp. 1460–1477.
Pipitone, J. et al. (2014) ‘Multi-atlas segmentation of the whole hippocampus and subfields using multiple automatically generated templates’, NeuroImage, 101, pp. 494–512.
Prasad, S. et al. (2022) ‘Early onset of Parkinson’s disease in India: Complicating the conundrum’, Parkinsonism & related disorders, 105, pp. 111–113.
Theaud, G. et al. (no date) ‘TractoFlow: A robust, efficient and reproducible diffusion MRI pipeline leveraging Nextflow & Singularity’. Available at: https://doi.org/10.1101/631952.