Neural Modeling and Imaging of Speech

Boston University (Charles River Campus), Boston MA

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): The primary aims of this research project are (1) to further refine and test a neural model of the brain interactions underlying learning and production of speech sounds called the DIVA model, (2) to develop and test a novel computational model of spasmodic dysphonia based on this model. These aims will be pursued by combining neuroimaging, computational modeling, and large-sample data analysis to investigate fluent and disordered speech production. The proposed studies focus on the interactions between cortical (motor, auditory, and somatosensory) and subcortical (cerebellum, basal ganglia, and thalamus) structures during feedforward and feedback speech motor control processes. The findings from this research will be integrated into a single, improved version of the DIVA model, ensuring a unified theoretical account. This model will be implemented in computer software and simulated on the same tasks as speakers in associated functional magnetic resonance imaging (fMRI) experiments. Primary Aim 1 is addressed by two studies that seek to improve the mapping between functional modules in the DIVA model and neural substrates. In the first study, functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI) will be used to differentiate the contributions of distinct cerebellar regions to speech motor control. Findings will be used to improve the representation of cerebellar regions in the DIVA model. In the second study, functional and anatomical data pooled from several previous studies will be used to provide a highly reliable assessment of brain regions involved in different aspects of speech as well as the variability in these regions across the population. This information will be used to improve neuroanatomical localization of speech mechanisms in the DIVA model and to create a neuro-mechanistic atlas that relates speech motor processes to precisely defined anatomical locations. Primary Aim 2 is addressed by two studies designed to improve our understanding of the neural mechanisms underlying the voicing disorder spasmodic dysphonia (SD). First, a model of vocal fold dynamics will be incorporated within the DIVA model, and simulations of voicing and stress control will be performed to verify model performance. Second, fMRI and DTI will be used in combination with the SD model to test several hypotheses concerning the primary pathophysiology in SD and differentiate it from secondary consequences of the disorder. This project's approach of integrating large-scale neural network modeling and neuroimaging will provide a clearer, more mechanistic account of the neural processes underlying normal and dysfluent speech. The findings from this project will help guide speech language pathologists to make more informed decisions about possible therapies or drug treatments to apply to patients with SD or other speech motor disorders and, in the longer term, guide development of new therapies and treatments for these patients.