Distinct subgroup-level functional connectivity in stuttering and non-stuttering children
Poster No:
1054
Submission Type:
Abstract Submission
Authors:
Soo-Eun Chang1, Yanni Liu1, Fiona Höbler1, Hannah Becker1, Valeria Caruso1, Michael Angstadt1, Adriene Beltz1
Institutions:
1University of Michigan, Ann Arbor, MI
First Author:
Co-Author(s):
Introduction:
Developmental stuttering (DS) emerges during a heightened period of speech motor skill development, manifesting as blocks, prolongations, or repetitions on the initiation of speech sequences. DS has been modelled as a systems-level impairment1,2, with structural and functional anomalies identified in the basal ganglia-thalamo-cortical (BGTC) circuits that support planning and execution of speech motor sequences3,4,5. Yet, past studies have predominantly focused on select brain areas or connections through research with adult participants6,7, without leveraging analytical advances of network-based approaches. The aim of this study was to test hypothesized impairments in the speech network's motor and planning circuits among children who stutter, using the Group Iterative Multiple Model Estimation (GIMME)8 that allows deriving of group- as well as individual-level connectivity measures. We hypothesized that functional connectivity patterns detected with GIMME would show distinct subgroup-level network differences between stuttering and control groups in the (i) planning loop (ii) motor loop (iii) broader network defined in an established neurocomputational model of speech production (Gradient Order Directions Into Velocities of Articulators, GODIVA )9,10, encompassing feedback and feedforward structures.
Methods:
We used confirmatory subgrouping (CS-GIMME)11,12, a recent extension of GIMME, to estimate subgroup-level connections for priori known groups (stuttering, control). Connectivity results are derived at the group as well as at individual level, which allows examining subject-specific heterogeneity in connectivity. CS-GIMME can detect paths between nodes ("edges") that are consistently present for individuals within stuttering and control groups, thus facilitating our interpretation of the heterogeneous connectivity maps and allowing for subgroup-specific inferences.
Resting state fMRI (rsfMRI) data were acquired from 73 children who stutter (CWS) and 76 age- and gender-matched children who do not stutter (CNS) (mean age=72 ± 22 months, age range from 38-129 months, 34 CWS girls, 40 CNS girls). Stuttering severity (SSI) range was 2-37 (17.8±6.3) (very mild~very severe). Data were processed using standard methods in SPM12. Subjects were eligible to be included if they had at least 4 minutes of useable data (after motion censoring at FD>0.5mm) and a usable T1 image. Participant-specific time series (164 functional volumes) from 17 regions of interest (ROIs) were extracted. The ROIs and their locations were selected according to regions defined in the DIVA model (Tourville & Guenther, 2011). CS-GIMME was run using a threshold of 75% for group-level edges and 50% for subgroup-level edges.
Resting state fMRI (rsfMRI) data were acquired from 73 children who stutter (CWS) and 76 age- and gender-matched children who do not stutter (CNS) (mean age=72 ± 22 months, age range from 38-129 months, 34 CWS girls, 40 CNS girls). Stuttering severity (SSI) range was 2-37 (17.8±6.3) (very mild~very severe). Data were processed using standard methods in SPM12. Subjects were eligible to be included if they had at least 4 minutes of useable data (after motion censoring at FD>0.5mm) and a usable T1 image. Participant-specific time series (164 functional volumes) from 17 regions of interest (ROIs) were extracted. The ROIs and their locations were selected according to regions defined in the DIVA model (Tourville & Guenther, 2011). CS-GIMME was run using a threshold of 75% for group-level edges and 50% for subgroup-level edges.
Results:
Group differences in network density were observed in the posterior inferior frontal sulcus (pIFS) within the Planning loop (Fig 1): 1) control group showed greater connectivity between the left pIFS and the caudate; 2) CWS showed greater connectivity of the left pIFS with the ventral later thalamus (VL) within the Motor Loop. Overall, there was a greater number of connections within the planning loop structures for controls relative to CWS (Fig 2). No group differences were observed in network connectivity found for motor loop or among DIVA structures.
Conclusions:
These results show that CS-GIMME can derive functional connectivity results that differentiate stuttering from non-stuttering groups in pathways predicted by a neurocomputational model of speech processing (GODIVA, DIVA). The current findings suggest that stuttering may be associated with impaired planning of speech sounds in their sequential order. We plan to further apply CS-GIMME to examine persistent vs. recovered groups within the stuttering group. In future research, we will further apply GIMME to derive data-driven subgroups within the group of children who stutter to examine whether this method can help predict specific subtypes, or eventual persistence and recovery in developmental stuttering.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism)
Language:
Speech Production 1
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling 2
Motor Behavior:
Motor Planning and Execution
Keywords:
Basal Ganglia
DISORDERS
FUNCTIONAL MRI
Pediatric Disorders
1|2Indicates the priority used for review
Provide references using author date format
Bohland, J.W. (2010), ‘Neural representations and mechanisms for the performance of simple speech sequences’, Journal of cognitive neuroscience, vol. 22, no. 7, pp. 1504-1529.
Cai, S. (2014), ‘Diffusion imaging of cerebral white matter in persons who stutter: evidence for network-level anomalies’, Frontiers in human neuroscience, vol. 8, no. 54, pp. 1-18.
Chang, S.E. (2020), ‘Involvement of the cortico-basal ganglia-thalamocortical loop in developmental stuttering’, Frontiers in Psychology, vol. 10, no. 3088, pp. 1-15.
Chow, H.M. (2023), ‘Brain developmental trajectories associated with childhood stuttering persistence and recovery’, Developmental Cognitive Neuroscience, vol. 60, no. 101224, pp. 1-14.
Gates, K.M. (2012), ‘Group search algorithm recovers effective connectivity maps for individuals in homogeneous and heterogeneous samples’, NeuroImage, vol. 63, no. 1, pp. 310-319.
Gates, K.M. (2017), ‘Unsupervised classification during time-series model building’, Multivariate behavioral research, vol. 52, no. 2, pp. 129-148.
Guenther, F.H. (2016), ‘Neural Control of Speech’, Cambridge, MA: MIT Press.
Henry, T.R. (2019), ‘Comparing directed functional connectivity between groups with confirmatory subgrouping GIMME’, NeuroImage, vol. 188, pp. 642-653.
Lu, C. (2010), ‘Altered effective connectivity and anomalous anatomy in the basal ganglia-thalamocortical circuit of stuttering speakers’, Cortex, vol. 46, no. 1, pp. 49-67.
Masapollo, M. (2021), ‘Behavioral and neural correlates of speech motor sequence learning in stuttering and neurotypical speakers: an fMRI investigation’, Neurobiology of Language, vol. 2, no. 1, pp. 106-137.
Tourville, J.A. (2019), ‘Functional parcellation of the speech production cortex’, Journal of Speech, Language, and Hearing Research, vol. 62, no. 8S, pp. 3055-3070.
Tourville, J.A. (2011), “The DIVA model: A neural theory of speech acquisition and production’, Language and cognitive processes, vol. 26, no. 7, pp. 952-981.