Speaker:  Vaibhav A. Diwadkar, Professor (https://wayne.edu/people/ax3112)

Dept. of Psychiatry & Behavioral Neurosciences, WSU

Time: Wednesday, September 4th, from 2:30 to 3:30 pm

Location for in-person participants: 1146 FAB 

Zoom link for online audience: https://wayne-edu.zoom.us/j/92845590121?pwd=CpRA5Wa5gzSMn2xiVkR2abD83O5nrH.1

 

Title: Unpacking the loss of coordination dynamics in the schizophrenia brain: Experiments on fMRI data

 

Abstract: Nothing about the human brain or about clinical conditions that emerge from it, is particularly straightforward (“If the human brain were so simple that we could understand it, we would be so simple that we couldn’t”: Emerson Pugh).” Schizophrenia is one of the most perplexing of these clinical conditions. The term’s roots are literal: From the nature of patient’s symptoms, Blueler and Kraepelin thought of schizophrenia as a kind of “splitting of the mind”. In the age of theoretical neurobiology and computational psychiatry, this metaphorical concept is now operationalized under the term “dys-connection” (Friston et al., 2016). In its essence, the dys-connection hypothesis suggests that schizophrenia’s clinical characteristics are fundamentally linked to the brain’s underlying molecular and neuronal pathophysiology. Thus, psychosis (like any normal or abnormal behavior) is expressed in contextually modulated changes at a system’s level, where the system may be microscopic (e.g., neuronal) or macroscopic (e.g., large scale human brain networks) in scale. Advances in in vivo functional magnetic resonance imaging (fMRI) now allow us to usefully characterize brain network function and dysfunction at the macroscopic scale (Diwadkar & Eickhoff, 2022); at this scale, brain regions are defined based on anatomy and/or function, and the regions aggregate across the brain’s underlying microstructure. Acquired imaging signals can be analyzed to understand functional relationships between defined regions. Accordingly, problems previously defined as purely “biological” in nature, can now be restated as computational, mathematical or biophysical.

 

Recent theoretical work identifies coordination dynamics as a central mechanism underpinning dynamically unfolding brain function (Bressler & Kelso, 2016). These dynamics arise from a combination of cooperative and competitive interactions between brain regions (Razi & Friston, 2016), and theoretical considerations suggest that coordination dynamics are likely to be aberrant in schizophrenia (Bressler, 2003), providing a particular manifestation of dys-connection. However, few (if any) studies have attempted to study coordination dynamics in the schizophrenia brain. Here, I will present our attempts to quantify and interpret the loss of coordination dynamics in schizophrenia. fMRI data were collected (78 total patients and typical controls) using a specifically curated associative learning paradigm (Martin et al., 2024) known to evoke large scale cross-cerebral dynamics (Meram et al., 2023). Next, task-evoked dynamics were captured using dynamic functional connectivity (DFC)(Hutchison et al., 2013): In each participant, and in each of 37 successive partially overlapping time windows, we estimated undirected functional connectivity (uFC, Pearson’s r, based on zero lag correlations between fMRI time series) across the full cerebral cross-correlation matrix (90 anatomically defined regions, C(90,2) = 4005 unique inter-regional relations). Each element of the 4005 unique cells represents the degree of coordinated activity between any pair of regions (i,j) in that time window. Next, for each cell (uFC_(i,j)) in each participant’s data, we formed a time series from the 37 successive Pearson’s r values. Each such time series represents the coordination dynamics for any two pairs (i,j) of regions across the task (henceforth, CD_(i,j)). We next clustered the superset of CD_(i,j) (312,390 across all 78 participants) based on similarities between all possible pairs (C(312390,2), i.e., ~4.88 × 10¹⁰ pairs), where similarities were estimated using dynamic time warping (DTW)(Berndt & Clifford, 1994). Next, we used k-medoids clustering (Kaufman & Rousseeuw, 1994) to partition the CD_(i,j) (10 partitions) before examining the partitions for coordination dynamics and membership.

 

Our results were clear. Partitions characterized by a high degree of coordination dynamics were dominated by data from typical controls. Brain regions involved in maintaining these dynamics included the dorsolateral prefrontal cortex, the parietal cortex, the medial temporal lobe and the thalamus. Conversely, partitions characterized by a low degree of coordination dynamics were dominated by data from schizophrenia patients, with brain regions diffusely represented.Our results are the first to demonstrate a loss of coordination dynamics in schizophrenia and have salient implications for understanding of this condition. I will discuss the import of our results as well as the utility of our pipeline for studying coordination dynamics across tasks and other clinical conditions.