A hitchhiker’s guide to working with large, open-source neuroimaging datasets

Decision Models and Technology Can Help Psychiatry Develop Biomarkers

“…Meanwhile, predictive biomarker studies can overfit the parameters of their model to a given dataset; therefore, even though a predictive biomarker might explain a large amount of the variance, this model is useless in a novel clinical population. Although cross-validation techniques are meant to help minimize the likelihood of overfitting ( 25 , 37 ), many datasets are unique, so cross-validating on an independent but similarly unique dataset does not truly demonstrate generalizability or resolve the overfitting problem ( 38 ). Grounding biomarker studies in clinical practice and making utility within a decision model a necessary component of biomarker development, therefore, might prove helpful.…”

“…The relatively nascent field of Deep Phenotyping aims to collect data for large, longitudinal samples using standardized and rigorous procedures ( 87 ). Multiple on-going, large-scale, necessarily collaborative efforts are seeking to provide deep phenotypes ( 88 ) that span genetic and epigenetic data to brain imaging to digitized behavioral and online data ( 89 , 90 ). Together these data seek to measure—as much as possible—an individual's biologic, social, and psychologic profile.…”

Virtual Connectomic Datasets in Alzheimer’s Disease and Aging Using Whole-Brain Network Dynamics Modelling

“…One of the greatest challenges today is to develop approaches allowing the useful exploitation of large-scale datasets in biomedical research in general ( Margolis et al, 2014 ) and neuroscience and neuroimaging in particular ( Van Horn and Toga, 2014 ). Progress in this direction is made possible by the increasing availability of large public datasets in the domain of connectomics ( Van Essen et al, 2013 ; Poldrack and Gorgolewski, 2014 ; Horien et al, 2021 ). This is true, in particular, for research in Alzheimer’s disease (AD), in which, despite decades of massive investment and a daunting literature on the topic, the partial and, sometimes contradictory nature of the reported results ( Patterson, 2018 ) still prevents a complete understanding of the factors governing the progression of the disease ( Braak and Braak, 1991 ; Braak et al, 2006 ; Komarova and Thalhauser, 2011 ; Henstridge et al, 2019 ) or of the diversity of cognitive deficits observed in different subjects ( Iacono et al, 2009 ; Mungas et al, 2010 ; Allen et al, 2016 ).…”

A protocol for working with open-source neuroimaging datasets

“…Finally, we give instructions for data and result sharing on public repositories and preprint services. For complete details on the use and execution of this profile, please refer to Horien et al. (2021) .…”

“…For complete details on the use and execution of this profile, please refer to Horien et al. (2021) .…”