Cortical similarities in psychiatric and mood disorders identified in federated VBM analysis via COINSTAC

Abstract

Structural neuroimaging studies have identified a combination of shared and disorder-specific patterns of gray matter (GM) deficits across psychiatric disorders. Pooling large data allows for examination of a possible common neuroanatomical basis that may identify a certain vulnerability for mental illness. Large-scale collaborative research is already facilitated by data repositories, institutionally supported databases, and data archives. However, these data-sharing methodologies can suffer from significant barriers. Federated approaches augment these approaches by enabling access or more sophisticated, shareable and scaled-up analyses of large-scale data. We examined GM alterations using Collaborative Informatics and Neuroimaging Suite Toolkit for Anonymous Computation, an open-source, decentralized analysis application. Through federated analysis of eight sites, we identified significant overlap in the GM patterns (n = 4,102) of individuals with schizophrenia, major depressive disorder, and autism spectrum disorder. These results show cortical and subcortical regions that may indicate a shared vulnerability to psychiatric disorders.

Keywords: COINSTAC; PTSD; decentralized; federated analysis; gray matter; mild cognitive impairment; mood disorders; psychiatric disorders; regression; transdiagnostic.

Figure 1

Gray matter alteration in schizophrenia Alterations in the gray matter of individuals with schizophrenia when compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 2

Gray matter alteration in bipolar disorder Alterations in the gray matter of individuals with bipolar disorder when compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 3

Gray matter alteration in major depressive disorder Alterations in the gray matter of individuals with major depressive disorder when compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 4

Gray matter alteration in post-traumatic stress disorder Alterations in the gray matter of individuals with post-traumatic stress disorder when compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 5

Gray matter alteration in autism spectrum disorder Alterations in the gray matter of individuals with autism spectrum disorder when compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 6

Gray matter alteration in mild cognitive impairment Alterations in the gray matter of individuals with mild cognitive impairment when compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 7

Gray matter alteration in spectrum Alterations in the gray matter of individuals identified with a spectrum disorder when compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 8

Gray matter alteration in unaffected first-degree relatives Alterations in the gray matter of unaffected relatives compared to healthy volunteers. The colormap shows the -log(p) value; warm colors are used for positive values and cool colors for negative values.

Figure 9

Heatmap of spatial correlations Correlations for each pair of diagnostic group comparisons. Rel, unaffected relatives; PTSD, post-traumatic stress disorder; MCI, mild cognitive impairment; BP, bipolar disorder; ASD, autism spectrum disorder; MD, major depressive disorder; SZ, schizophrenia; Spect, spectrum.

Figure 10

Heatmap of spatial correlations with MCI correlated out of comparisons Correlations for each pair of diagnostic group comparisons. Rel, unaffected relatives; PTSD, post-traumatic stress disorder; MCI, mild cognitive impairment; BP, bipolar disorder; ASD, autism spectrum disorder; MD, major depressive disorder; SZ, schizophrenia; Spect, spectrum.

Figure 11

The COINSTAC architecture A diagram of the COINSTAC architecture, including both client nodes and remote services. The COINSTAC application uses standard web communication protocols to transfer (meta-)data between each client node and a publicly accessible remote node via an API gateway. Client nodes and the remote node run computations in a pipeline, with client nodes sending meta-data derived from local data and the remote node aggregating and running required computations, sending results back to client nodes for either final output or iteration if required. Pipeline results are stored in a noSQL data store provided by MongoDB.