Harmony: generating a large-scale connectome dataset for mental health research
Poster No:
2228
Submission Type:
Abstract Submission
Authors:
Petra Lenzini1, Leonardo Tozzi2, Lexi (Xiaoke) Luo3, Jenna Jubeir4, Artemis Zavaliangos-Petropulu5, Jiahe Zhang6, Timothy Lyons7, Anna Broken7, Adam Pines8, Susan Whitfield Gabrieli6, Katherine Narr9, Leanne Williams7, Yvette Sheline10, Janine Bijsterbosch1
Institutions:
1Washington University in St Louis, St Louis, MO, 2Department of Psychiatry and Behavioral Sciences, Stanford University, Palo Alto, CA, 3Washington University in Saint Louis, St Louis, MO, 4Department of Psychiatry & Behavioral Sciences, Stanford University School of Medicine, Stanford, Palo Alto, CA, 5Department of Neurology, Geffen School of Medicine at the University of California, Los Angeles, CA, 6Department of Psychology, Northeastern University and Department of Psychiatry, MGH, Boston, MA, 7Stanford University School of Medicine, Palo Alto, CA, 8Psychiatry and Behavioral Sciences, Stanford School of Medicine, Stanford University, Palo Alto, CA, 9University of California Los Angeles, Los Angeles, CA, 10University of Pennsylvania Perelman School of Medicine, Philadelphia, PA
First Author:
Co-Author(s):
Jenna Jubeir
Department of Psychiatry & Behavioral Sciences, Stanford University School of Medicine, Stanford
Palo Alto, CA
Department of Psychiatry & Behavioral Sciences, Stanford University School of Medicine, Stanford
Palo Alto, CA
Artemis Zavaliangos-Petropulu
Department of Neurology, Geffen School of Medicine at the University of California
Los Angeles, CA
Department of Neurology, Geffen School of Medicine at the University of California
Los Angeles, CA
Jiahe Zhang
Department of Psychology, Northeastern University and Department of Psychiatry, MGH
Boston, MA
Department of Psychology, Northeastern University and Department of Psychiatry, MGH
Boston, MA
Adam Pines
Psychiatry and Behavioral Sciences, Stanford School of Medicine, Stanford University
Palo Alto, CA
Psychiatry and Behavioral Sciences, Stanford School of Medicine, Stanford University
Palo Alto, CA
Susan Whitfield Gabrieli
Department of Psychology, Northeastern University and Department of Psychiatry, MGH
Boston, MA
Department of Psychology, Northeastern University and Department of Psychiatry, MGH
Boston, MA
Introduction:
Following the success of the initial Human Connectome Project (HCP) dataset, the NIH has funded several follow-up Connectomes Related to Human Disease (CRHD) studies. These CRHD studies were intended to adopt high-quality HCP data acquisition and processing techniques in disease-specific cohorts. Four CRHD studies collected data from mental health cohorts suffering from mood and anxiety disorders (Tozzi et al., 2021), namely: Dimensional Connectomics of Anxious Misery (HCP-ANXPE; PI Sheline), Human Connectome Project for Disordered Emotional States (HCP-DES; PI Williams; Tozzi et al., 2020), Perturbation of the treatment resistant depression connectome by fast-acting therapies (HCP-MDD; PI Narr), and Connectomes related to anxiety and depression in adolescents (HCP-BANDA; PI Whitfield-Gabrieli). However, recent work has shown that larger sample sizes than each of these individual studies (Ncases<250) contain might be required to identify robust brain-behavior associations (Marek et al., 2022). The goal of the Harmony project is to combine these studies and supplement controls from other cohorts to generate a harmonized large-scale connectome dataset for mental health research.
Methods:
The combined Harmony sample will include N=1,540 participants, covering 770 cases and 770 matched controls (Table 1). 1,736 non-imaging variables spanning 52 instruments in 7 domains were harmonized and centrally annotated. Quality control based on the systematic inspection of variable distributions was performed to identify and correct source harmonization errors. All harmonization steps are publicly available, version controlled, and annotated to track provenance to maximize reproducibility.
Preprocessing of neuroimaging datasets consists of the HCP minimal processing pipeline (Glasser et al., 2013) with MSMAll surface-based alignment (Robinson et al., 2018) and ICA-FIX cleanup (Salimi-Khorshidi et al., 2014) as implemented in Qunex (Ji et al., 2022). The resulting processed data are in CIFTI format using a 32k mesh matching HCP-YA data. Preprocessed neuroimaging data will be shared through the NIMH Data Archive (NDA).
Multimodal Imaging Derived Phenotypes (IDPs) will be extracted from common pipelines inspired by HCP-YA and UK Biobank datasets. Resting state IDPs will include amplitudes, full correlation matrices, and partial correlation matrices from the following parcellations: HCP-MMP1.0 (Glasser et al., 2016), Independent Component Analysis combined with dual regression, and PROFUMO (Harrison et al., 2020). Diffusion IDPs will include fractional anisotropy (FA), mean diffusivity (MD), and NODDI indices of neurite density index (IcVF), orientation dispersion index (ODI), and isotropic volume fraction (IsoVF) (Zhang et al., 2012). Structural IDPs will include cortical area, thickness, volume, and myelination (T1/T2) from HCP-MMP1.0 and DKT cortical parcellations and from ASEG and Harvard-Oxford subcortical segmentations. All IDPs will be harmonized using longitudinal COMBAT to account for site effects. Harmonized IDPs will be shared.
Preprocessing of neuroimaging datasets consists of the HCP minimal processing pipeline (Glasser et al., 2013) with MSMAll surface-based alignment (Robinson et al., 2018) and ICA-FIX cleanup (Salimi-Khorshidi et al., 2014) as implemented in Qunex (Ji et al., 2022). The resulting processed data are in CIFTI format using a 32k mesh matching HCP-YA data. Preprocessed neuroimaging data will be shared through the NIMH Data Archive (NDA).
Multimodal Imaging Derived Phenotypes (IDPs) will be extracted from common pipelines inspired by HCP-YA and UK Biobank datasets. Resting state IDPs will include amplitudes, full correlation matrices, and partial correlation matrices from the following parcellations: HCP-MMP1.0 (Glasser et al., 2016), Independent Component Analysis combined with dual regression, and PROFUMO (Harrison et al., 2020). Diffusion IDPs will include fractional anisotropy (FA), mean diffusivity (MD), and NODDI indices of neurite density index (IcVF), orientation dispersion index (ODI), and isotropic volume fraction (IsoVF) (Zhang et al., 2012). Structural IDPs will include cortical area, thickness, volume, and myelination (T1/T2) from HCP-MMP1.0 and DKT cortical parcellations and from ASEG and Harvard-Oxford subcortical segmentations. All IDPs will be harmonized using longitudinal COMBAT to account for site effects. Harmonized IDPs will be shared.
·Demographic overview of the Harmony dataset.
Results:
At the time of abstract submission, non-imaging instruments and variables released through the NDA were cleaned and their metadata standardized. Furthermore preprocessing of neuroimaging data for MDD and BANDA has been completed and ANXPE data have been staged for preprocessing. See Figure 1 for a timeline overview. Community feedback on the Harmony processing and data sharing plans is encouraged. Feedback promoting the use of annotation standards which facilitate the steady transition to community use of persistent ontological identifiers is also encouraged.
·Timeline for the Harmony dataset.
Conclusions:
The large-scale connectome-quality Harmony dataset will be a valuable community data resource for research into mental health. In particular, the Harmony dataset will facilitate the clinical translation of advances in methodological development and brain organization insights that have and continue to result from healthy cohorts such as the HCP-YA.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 2
Neuroinformatics and Data Sharing:
Databasing and Data Sharing 1
Keywords:
Anxiety
Other - Depression, Harmonization, Data Sharing
1|2Indicates the priority used for review
Provide references using author date format
Glasser, M. F., et al (2013). The minimal preprocessing pipelines for the Human Connectome Project. NeuroImage, 80, 105–124. https://doi.org/10.1016/j.neuroimage.2013.04.127
Harrison, S. J., et al (2020). Modelling subject variability in the spatial and temporal characteristics of functional modes. NeuroImage, 222, 117226. https://doi.org/10.1016/j.neuroimage.2020.117226
Ji, J. L., Demšar, et al (2022). QuNex – An Integrative Platform for Reproducible Neuroimaging Analytics. In bioRxiv. https://doi.org/10.1101/2022.06.03.494750
Marek, S., et al (2022). Reproducible brain-wide association studies require thousands of individuals. Nature, 603(7902), 654–660. https://doi.org/10.1038/s41586-022-04492-9
Robinson, E. C., et al (2018). Multimodal surface matching with higher-order smoothness constraints. NeuroImage, 167, 453–465. https://doi.org/10.1016/j.neuroimage.2017.10.037
Salimi-Khorshidi, G., et al (2014). Automatic denoising of functional MRI data: combining independent component analysis and hierarchical fusion of classifiers. NeuroImage, 90, 449–468. https://doi.org/10.1016/j.neuroimage.2013.11.046
Tozzi, L., et al (2021). Convergence, preliminary findings and future directions across the four human connectome projects investigating mood and anxiety disorders. NeuroImage, 245, 118694. https://doi.org/10.1016/j.neuroimage.2021.118694
Tozzi, L., et al. (2020). The human connectome project for disordered emotional states: Protocol and rationale for a research domain criteria study of brain connectivity in young adult anxiety and depression. NeuroImage, 214, 116715. https://doi.org/10.1016/j.neuroimage.2020.116715
Zhang, H., et al (2012). NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. NeuroImage, 61(4), 1000–1016. https://doi.org/10.1016/j.neuroimage.2012.03.072