Table of contents

1. Neuromark
   1. Summary
2. NDAR

Neuromark

Clinical similarities among brain disorders have been recognized for many decades. Although some brain disorders are now recognized as distinct diseases, they share overlapping genetic etiology, common abnormalities in brain function and structure, and similar cognitive impairments, confounding the diagnosis and treatment plans. Neuroimaging, a more specific and biologically based approach for detecting brain changes, has advanced our knowledge of disease-related brain dysfunction and provided reliable biomarkers for clinical usage in some cases. However, previous studies typically focus on one or two disorders only, ignoring the study of overlapping and unique brain abnormalities among brain disorders. Characterizing brain changes across disorders using neuroimaging approaches and understanding the connections and relationships between them may shed light on etiologies and aid in the deconstruction of psychiatric illnesses. Indeed, how to compare biomarkers across multi-disorders and how to replicate biomarkers are two critical issues which attract increasing consideration in the neuroimaging research field. Region of interest (ROI) based analysis and independent component analysis (ICA) are the two most common approaches for exploring the functional organization of the brain. While ROI based methods require fixed brain regions according to prior experiences or knowledge from in-hand data, ICA, a data-driven method with an assumption of independence between identified components, is capable of capturing functional features retaining individual variability. Our previous studies along with others have found that a set of features derived from ICA can serve as potential biomarkers of several brain disorders, with the ability to predict disease traits and cognitive declines. However, unlike univariate methods, comparing results obtained from different ICA runs is not so straightforward due to the difference in the identified target components and their arrangements, which greatly raises difficulty on examining results from multiple studies and also decreases the possibility for the replication of existing results. In this project, we aim to launch a novel unified ICA framework which can be used to link multiple diseases, datasets, and studies. The framework is capable of generating more accurate individual-level functional networks as well as to identify the correspondence of networks across subjects/datasets/studies, also useful to evaluate replication. Based on this framework, comparable ICA features, such as the spatial maps within functional networks and functional (among) network connectivity, will be extracted from each subject respectively. We will employ this framework using a database created from a large sample healthy controls and individuals with different brain disorders (schizophrenia, autism, attention-deficit/hyperactivity disorder, and major depression disorder). This database will be formed by combining data collected from multiple open-source and our own datasets. It is expected that by using this robust and generalized framework, disease overlap and divergence of various ICA features can be quantified and summarized. The results might help to provide credible evidence regarding the neural basis contributing to the overlapping symptoms in brain disorders and thus contribute to the understanding of the specificity of brain disorders as well as their inter-relationships (such as continuously changing, distinct, or common) among them.

References

[1] Y. Du et al., “NeuroMark: an automated and adaptive ICA based pipeline to identify reproducible fMRI markers of brain disorders,” NeuroImage: Clinical, p. 102375, Aug. 2020, doi: 10.1016/j.nicl.2020.102375.

Summary

Collaborative spreadsheet: https://drive.google.com/file/d/126Su7fBqaVAjefCe4s4Prk9qk_zdzsVy/view?usp=sharing

NDAR

Follow the steps below to download data from NDAR.

• Go to https://nda.nih.gov/general-query.html?q=query=collections%20~and~%20orderBy=id%20~and~%20orderDirection=Ascending
• Select what you like to download
• Add to filter (takes long time)
• Create a package (also takes long time)
• All data in the package can be downloaded using this Python package: https://github.com/NDAR/nda-tools
• The preferred method for downloading scan data files is as follows:
  • Download a package without associated data, e.g. downloadcmd ` -dp`
  • Then extract the s3 locations from the downloaded image03.txt file, e.g. remain.txt

$ head remain.txt
s3://**************/submission_13124/NDARINVRP3R4YCP_baselineYear1Arm1_ABCD-rsfMRI_**************.tgz
s3://**************/submission_13124/NDARINVTAX3MN8C_baselineYear1Arm1_ABCD-rsfMRI_**************.tgz
s3://**************/submission_13124/NDARINVJ2ENP1ZK_baselineYear1Arm1_ABCD-rsfMRI_**************.tgz
s3://**************/submission_13124/NDARINVUKPZU1JW_baselineYear1Arm1_ABCD-rsfMRI_**************.tgz
s3://**************/submission_13124/NDARINV5VGKMHCR_baselineYear1Arm1_ABCD-rsfMRI_**************.tgz

• Finally, download through the list remain.txt, e.g. downloadcmd -t remain.txt -d image_files/