Dynamic functional connectivity analysis reveals transiently increased segregation in patients with severe stroke

Abstract

Background and Purpose To explore the whole-brain dynamic functional network connectivity patterns in acute ischemic stroke (AIS) patients and their relation to stroke severity in the short and long term. Methods We investigated large-scale dynamic functional network connectivity of 41 AIS patients two to five days after symptom onset. Re-occurring dynamic connectivity configurations were obtained using a sliding window approach and k-means clustering. We evaluated differences in dynamic patterns between three NIHSS-stroke severity defined groups (mildly, moderately, and severely affected patients). Furthermore, we established correlation analyses between dynamic connectivity estimates and AIS severity as well as neurological recovery within the first 90 days after stroke. Finally, we built Bayesian hierarchical models to predict acute ischemic stroke severity and examine the inter-relation of dynamic connectivity and clinical measures, with an emphasis on white matter hyperintensity lesion load. Results We identified three distinct dynamic connectivity configurations in the early post-acute stroke phase. More severely affected patients (NIHSS 10-21) spent significantly more time in a highly segregated dynamic connectivity configuration that was characterized by particularly strong connectivity (three-level ANOVA: p<0.05, post hoc t-tests: p<0.05, FDR-corrected for multiple comparisons). Recovery, as indexed by the realized change of the NIHSS over time, was significantly linked to the acute dynamic connectivity between bilateral intraparietal lobule and left angular gyrus (Pearson's r=-0.68, p<0.05, FDR-corrected). Increasing dwell times, particularly those in a very segregated connectivity configuration, predicted higher acute stroke severity in our Bayesian modelling framework. Conclusions Our findings demonstrate transiently increased segregation between multiple functional domains in case of severe AIS. Dynamic connectivity involving default mode network components significantly correlated with recovery in the first three months post-stroke.

Dynamic connectivity predicts acute motor impairment and recovery post-stroke

“…17 Furthermore, subsets of the data were previously analyzed with respect to the relationship of acute motor symptoms and connectivity. 8 , 42 Importantly, these previous analyses considered data from the acute phase only and all analyses presented in this paper are new, hence there is no overlap.…”

“…Previous studies of dynamic connectivity in independent acute ischemic stroke cohorts have already uncovered links between functional integration and segregation and motor performance: lower motor domain integration in case of severe compared to moderate hand motor impairments 17 and increased whole-brain segregation (i.e., decreased integration) in case of a high stroke severity. 42 Interestingly, in healthy subjects with cast-induced motor inactivity of the upper limb, motor network topology also turned into a more segregated state. 43 We here did not observe a clear correlation between fraction times in integrated state 3 and acute motor impairment across all subjects.…”

Neuropsychiatric disease classification using functional connectomics - results of the connectomics in neuroimaging transfer learning challenge

“…Functional connectomics, or the study of whole-brain synchronization maps, has become of increasing interest to the neuroscientific community in recent years. For example, functional connectomics has provided valuable insight into human cognition (Mill et al (2017); Shine et al (2016)), the systemlevel organization of the brain over development and aging (Somerville et al (2018); Kaiser (2017)), and whole-brain functional alterations in disease or injury (D'Souza et al (2019);D'Souza et al (2020); Venkataraman et al (2012Venkataraman et al ( , 2013bVenkataraman et al ( , 2016Venkataraman et al ( , 2013a; Ktena et al (2019); Bonkhoff et al (2020)). Large, open-source initiatives, such as the Human Connectome Project (Essen et al (2012)) and the Autism Brain Imaging Data Exchange (Di Martino et al (2014)), have spurred the development of new and increasingly powerful machine learning strategies to capitalize on these resources.…”

Increased prevalence of a frontoparietal brain state is associated with better motor recovery after stroke affecting dominant-hand corticospinal tract

“…This approach yields FC networks that fluctuate over time, with a temporal resolution proportional to the size of the window; about 30-60 seconds (Savva et al, 2019). In stroke populations, dynamic FC studies have demonstrated stroke-related differences in temporal configurations of motor networks (Bonkhoff et al, 2020a) and participation in connectivity states that varies with severity (Bonkhoff et al, 2020b).…”