Hyperactive sensorimotor cortex during voice perception in spasmodic dysphonia

Yuji Kanazawa,Shinya Hiroshiba,Yo Kishimoto,Kimihiro Nakamura,Toshihiko Aso,Koichi Omori,Ichiro Tateya,Toru Ishii,Tetsuji Sanuki

doi:10.1038/s41598-020-73450-0

Abstract

Spasmodic dysphonia (SD) is characterized by an involuntary laryngeal muscle spasm during vocalization. Previous studies measured brain activation during voice production and suggested that SD arises from abnormal sensorimotor integration involving the sensorimotor cortex. However, it remains unclear whether this abnormal sensorimotor activation merely reflects neural activation produced by abnormal vocalization. To identify the specific neural correlates of SD, we used a sound discrimination task without overt vocalization to compare neural activation between 11 patients with SD and healthy participants. Participants underwent functional MRI during a two-alternative judgment task for auditory stimuli, which could be modal or falsetto voice. Since vocalization in falsetto is intact in SD, we predicted that neural activation during speech perception would differ between the two groups only for modal voice and not for falsetto voice. Group-by-stimulus interaction was observed in the left sensorimotor cortex and thalamus, suggesting that voice perception activates different neural systems between the two groups. Moreover, the sensorimotor signals positively correlated with disease severity of SD, and classified the two groups with 73% accuracy in linear discriminant analysis. Thus, the sensorimotor cortex and thalamus play a central role in SD pathophysiology and sensorimotor signals can be a new biomarker for SD diagnosis.

Highlights

Because Spasmodic dysphonia (SD) patients, unlike healthy controls, should have much difficulty in vocalization, i.e., the act of voice production in itself should be much more limited and effortful for SD patients than for controls, which engages the motor, auditory and cognitive systems differently in the two groups and thereby creates a large confounding factor in between-group comparison analyses
We investigated whether fMRI signals during speech perception can serve as a specific biomarker reflecting the pathophysiology of SD
The present study used fMRI to isolate specific neural correlates of SD during speech perception

Summary

Introduction

Because SD patients, unlike healthy controls, should have much difficulty in vocalization, i.e., the act of voice production in itself should be much more limited and effortful for SD patients than for controls, which engages the motor, auditory and cognitive systems differently in the two groups and thereby creates a large confounding factor in between-group comparison analyses This may explain the fact that the reported patterns of neural activation are rather inconsistent across those previous studies. The authors observed increased activation in the primary sensorimotor cortex, insula, superior temporal gyrus, basal ganglia, thalamus and cerebellum in SD patients as compared to controls While this experimental manipulation largely allowed matching of the amount of vocal outputs between SD patients and controls, the observed neural effects may reflect a strategic recourse to other neurocognitive resources, because healthy volunteers would exert highly unnatural and effortful control over the normal speech production system. We assessed the diagnostic potential of the same fMRI data in distinguishing SD from healthy controls by using machine learning methods

Methods

Results

Conclusion