Format

Send to

Choose Destination
Brain. 2019 Jan 1;142(1):70-79. doi: 10.1093/brain/awy292.

Network localization of heterogeneous neuroimaging findings.

Darby RR1,2,3, Joutsa J2,4,5,6, Fox MD2,3,4.

Author information

1
Department of Neurology, Vanderbilt University Medical Center, Nashville, TN, USA.
2
Berenson-Allen Center for Noninvasive Brain Stimulation, Department of Neurology, Beth Israel Deaconess Medical Center, Harvard Medical Center, Boston, MA, USA.
3
Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.
4
Athinoula A. Martinos Centre for Biomedical Imaging, Massachusett General Hospital, Harvard Medical School, Charlestown, MA, USA.
5
Department of Neurology, University of Turku, Turku, Finland.
6
Division of Clinical Neurosciences, Turku University Hospital, Turku, Finland.

Abstract

Studies of the same disease often implicate different brain regions, contributing to a perceived reproducibility crisis in neuroimaging. Here, we leverage the normative human brain connectome to test whether seemingly heterogeneous neuroimaging findings localize to connected brain networks. We use neurodegenerative disease, and specifically Alzheimer's disease, as our example as it is one of the diseases that has been studied the most using neuroimaging. First, we show that neuroimaging findings in Alzheimer's disease occur in different brain regions across different studies but localize to the same functionally connected brain network. Second, we show that neuroimaging findings across different neurodegenerative diseases (Alzheimer's disease, frontotemporal dementia, corticobasal syndrome, and progressive non-fluent aphasia) localize to different disease-specific brain networks. Finally, we show that neuroimaging findings for a specific symptom within a disease (delusions in Alzheimer's disease) localize to a symptom-specific brain network. Our results suggest that neuroimaging studies that appear poorly reproducible may identify different regions within the same connected brain network. Human connectome data can be used to link heterogeneous neuroimaging findings to common neuroanatomy, improving localization of neuropsychiatric diseases and symptoms.

PMID:
30551186
PMCID:
PMC6308311
[Available on 2020-01-01]
DOI:
10.1093/brain/awy292

Supplemental Content

Full text links

Icon for Silverchair Information Systems
Loading ...
Support Center