|
|
|
12/30/15 |
       
|
Detection of Simulated Vocal Dysfunctions Using Complex sEMG Patterns by Luis Rivera and Nicholas Smith Introduction Symptoms of voice disorder may range from slight hoarseness to complete loss of voice; from modest vocal effort to uncomfortable neck pain. But even minor symptoms may still impact personal and especially professional lives. While early detection and diagnosis can ameliorate that effect, to date, we are still largely missing reliable and valid data to help us better screen for voice disorders. In our previous study, we started to address this gap in research by introducing an ambulatory voice monitoring system using surface electromyography (sEMG) and a robust algorithm (HiGUSSS) for pattern recognition of vocal gestures. Here, we expand on that work by further analyzing a larger set of simulated vocal dysfunctions. Our goal is to demonstrate that such a system has the potential to recognize and detect real vocal dysfunctions from multiple individuals with high accuracy under both intra and intersubject conditions. The proposed system relies on four sEMG channels to simultaneously process various patterns of sEMG activation in the search for maladaptive laryngeal activity that may lead to voice disorders. In the results presented here [1], our pattern recognition algorithm detected from two to ten different classes of sEMG patterns of muscle activation with an accuracy as high as 99%, depending on the subject and the testing conditions. Proposed Method: This work [1] uses a multi-channel version of the HiGUSSS method presented in [3]. The channel vectors are then averaged in order to form a single confidence feature vector to serve as the input to the Multi-Class SVM. Experiments and Results: This work drastically expanded on the testing and validation of the system proposed in [2] by: 1) more than doubling the number of test subjects (ten) with different ages and genders; 2) adding new groups of different gestures with both similar and distinct patterns representing normal and simulated dysfunctional conditions; 3) testing a larger number (ten) of vocal gestures; and 4) creating different test scenarios involving intra and inter-subject cases. The results show that despite the complexity of the muscle groups on the neck, meaningful detection of vocal dysfunctions through the recognition of sEMG signals is possible, at high levels of accuracy.
Figure 1. Muscle groups on the human neck: diagram, actual view of electrode placement, and actual view with bandage applied.
Figure 2. Means and standard deviations of the classification accuracies per gesture, over all ten subjects. Ten gestures considered: /a/, /a/ pressed, /u/, /u/ pressed, /i/, /i/ pressed, /t/, /s/, cough, and throat clear.
Figure 3. Means and standard deviations of the classification accuracies per subject, over all ten gestures. Ten gestures considered: /a/, /a/ pressed, /u/, /u/ pressed, /i/, /i/ pressed, /t/, /s/, cough, and throat clear.
Figure 4. Means and standard deviations of the classification accuracies over all subjects using normal vs. pressed gestures – i.e. simulated dysfunction.
Figure 5. Means and standard deviations of the classification accuracies per subject using normal vs. pressed gestures – i.e. simulated dysfunction.
Figure 6. Comparison between the average classification accuracies per subject, over all ten gestures, for each of the three classifier (HiGUSSS, MCSVM, and Distance). Set of all ten gestures considered: /a/, /a/ pressed, /u/, /u/ pressed, /i/, /i/ pressed, /t/, /s/, cough, and throat clear.
Conclusion: The results presented in [1] show that meaningful classification can be drawn from sEMG signals collected at the anterior neck. Overall, the HiGUSSS recognition system shows great promise in helping to better understand changes in vocal function that may be linked to voice disorders.
References
|
This site was last updated 12/30/15
