A multi-stage transfer learning strategy for diagnosing a class of rare laryngeal movement disorders.

BACKGROUND: It remains hard to directly apply deep learning-based methods to assist diagnosing essential tremor of voice (ETV) and abductor and adductor spasmodic dysphonia (ABSD and ADSD). One of the main challenges is that, as a class of rare laryngeal movement disorders (LMDs), there are limited available databases to be investigated. Another worthy explored research question is which above sub-disorder benefits most from diagnosis based on sustained phonations. The question is from the fact that sustained phonations can help detect pathological voice from healthy voice. METHOD: A transfer learning strategy is developed for LMD diagnosis with limited data, which consists of three fundamental parts. (1) An extra vocally healthy database from the International Dialects of English Archive (IDEA) is employed to pre-train a convolutional autoencoder. (2) The transferred proportion of the pre-trained encoder is explored. And its impact on LMD diagnosis is also evaluated, yielding a two-stage transfer model. (3) A third stage is designed following the initial two stages to embed information of pathological sustained phonation into the model. This stage verifies the different effects of applying sustained phonation on diagnosing the three sub-disorders, and helps boost the final diagnostic performance. RESULTS: The analysis in this study is based on clinician-labeled LMD data obtained from the Vanderbilt University Medical Center (VUMC). We find that diagnosing ETV shows sensitivity to sustained phonation within the current database. Meanwhile, the results show that the proposed multi-stage transfer learning strategy can produce (1) accuracy of 65.3% on classifying normal and other three sub-disorders all at once, (2) accuracy of 85.3% in differentiating normal, ABSD, and ETV, and (3) accuracy of 77.7% for normal, ADSD and ETV. These findings demonstrate the effectiveness of the proposed approach.

High-Speed Image Analysis Comparing Loading of Vocal Folds During Coughing and Phonation: A Case Study.

PURPOSE: Coughing is related to voice problems, since it involves firm glottal closure, fast glottal opening, and high subglottic pressure and flow rate. In this study, the glottal area variation and movements of laryngeal structures during coughing and phonation are compared. METHODS: High-speed laryngoscopy recordings were made of a normophonic male participant with a healthy larynx producing a neutral vowel and coughing. Oral air pressure was registered in a mouthpiece, through which an endoscope was inserted into the pharynx. Electroglottography, acoustic, and pressure signals were recorded simultaneously. The glottal width variation at the membranous and cartilaginous parts of the glottis was derived from the high-speed images, and the strong vibration of the false vocal folds was also registered. RESULTS: In coughing, compared to ordinary vowel phonation in nearly the same sound pressure level (93-94 dB6cm), the glottal width was 25% larger at the middle of the vocal folds, the maximum glottal opening velocity was 39% higher, and the maximum glottal width declination rate during glottal closing was up to three times higher. The maximum acceleration was 40% higher, and the maximum deceleration was 47% higher. Fundamental frequency f0 was the highest (ca. 400 Hz) at the beginning of the first phase of a typical coughing process. During the last part of the coughing process, f0 decreased from ca. 250 Hz to ca. 85 Hz at the phonation offset. CONCLUSIONS: The remarkable increase in maximum glottal width declination rate implies much higher vocal fold loading in coughing compared to phonation.

Conformable AlN Piezoelectric Sensors as a Non-invasive Approach for Swallowing Disorder Assessment.

Deglutition disorders (dysphagia) are common symptoms of a large number of diseases and can lead to severe deterioration of the patient's quality of life. The clinical evaluation of this problem involves an invasive screening, whose results are subjective and do not provide a precise and quantitative assessment. To overcome these issues, alternative possibilities based on wearable technologies have been proposed. We explore the use of ultrathin, compliant, and flexible piezoelectric patches that are able to convert the laryngeal movement into a well-defined electrical signal, with extremely low anatomical obstruction and high strain resolution. The sensor is based on an aluminum nitride thin film, grown on a soft Kapton substrate, integrated with an electrical charge amplifier and low-power, wireless connection to a smartphone. An ad-hoc designed laryngeal motion simulator (LMS), which is able to mimic the motions of the laryngeal prominence, was used to evaluate its performances. The physiological deglutition waveforms were then extrapolated on a healthy volunteer and compared with the sEMG (surface electromyography) of the submental muscles. Finally, different tests were conducted to assess the ability of the sensor to provide clinically relevant information. The reliability of these features permits an unbiased evaluation of the swallowing ability, paving the way to the creation of a system that is able to provide a point-of-care automatic, unobtrusive, and real-time extrapolation of the patient's swallowing quality even during normal behavior.

Evaluation of swallowing movement using ultrasonography.

The aim of this study is to develop an index to assess swallowing function by ultrasonography to evaluate the relationship between movements of the hyoid bone and the larynx while swallowing water. Forty-two younger participants (mean age, 20.3 ± 3.4 years) and 42 older participants (mean age, 75.1 ± 10.6 years) with normal swallowing function were included in the study. Movements of the hyoid bone and the larynx while swallowing 5 mL of water were observed using ultrasonography. Two-dimensional distances from the starting points of the hyoid bone and the larynx to their points of maximum movement were measured as displacements. The hyoid bone-laryngeal motion ratio was defined as the hyoid bone displacement divided by the laryngeal displacement. Parameters were compared among four groups: younger male, younger female, older male, and older female. The hyoid bone displacement differed significantly between the younger and older groups, and the laryngeal displacement differed significantly between age groups and sexes. The hyoid bone-laryngeal motion ratio was not significantly correlated with age, height, or body weight, and did not show a significant difference between the four groups. Thus, the hyoid bone-laryngeal motion ratio is an index that evaluates swallowing movement and is independent of physique and physiological changes associated with aging.

Office-based endoscopic botulinum toxin injection in laryngeal movement disorders.

INTRODUCTION: Botulinum toxin injection is widely used for the treatment of laryngeal movement disorders. Electromyography-guided percutaneous injection is the technique most commonly used to perform intralaryngeal botulinum toxin injection. OBJECTIVE: We describe an endoscopic approach for intralaryngeal botulinum toxin injection under local anaesthesia without using electromyography. TECHNIQUE: A flexible video-endoscope with an operating channel is used. After local anaesthesia of the larynx by instillation of lidocaine, a flexible needle is inserted into the operating channel in order to inject the desired dose of botulinum toxin into the vocal and/or vestibular folds. CONCLUSION: Endoscopic botulinum toxin injection under local anaesthesia is a reliable technique for the treatment of laryngeal movement disorders. It can be performed by any laryngologist without the need for electromyography. It is easy to perform for the operator and comfortable for the patient.

A new flexible piezoelectric pressure sensor array for the noninvasive detection of laryngeal movement during swallowing.

We tried to develop a new device to detect laryngeal movement noninvasively. We made small piezo pressure sensors (length, 1.5 mm, width, 7.0 mm), and five of these were lined up with 3.0-mm intervals and embedded in the middle of a palm-sized urethane resin sheet. This sheet was lightly attached to the ventral surface of the neck near the laryngeal prominence. The first and second peaks obtained from each sensor should correspond to the period when the larynx moves to the upper and lower positions during swallowing. The mean maximum rising velocities for men and women were about 0.08 and 0.11 m/s, respectively. Similarly, the mean maximum lowering velocities for men and women were about 0.09 and 0.11 m/s, respectively. The swallowing latencies for men and women were about 0.49 and 0.53 s, respectively. In conclusion, we succeeded in developing a new device, which will be useful in evaluating the swallowing function.

Sequential Tongue Pressure and Laryngeal Movement during Swallowing

OBJECTIVE: To test the repeatability and the reproducibility of a newly developed device which measures tongue pressure and laryngeal movement, to identify the range of tongue pressure and to reveal the relationships between tongue pressure and age, gender and diet formula. METHOD: One hundred five healthy subjects (50 males, 55 females, range 20 to 74 years) were recruited for the study. They had examinations of tongue pressure and laryngeal movement. The pressure was measured with two air-filled bulbs connected to a transducer. Laryngeal movement was measured with a strain gauge. The test was repeated three times with a two minute rest interval and monitored twice by the same investigator and once by another investigator. All subjects performed 10 times of swallowing: 5 times each of 3 ml liquid and dry swallowing. RESULTS: Intraclass correlation coefficient (ICC) for both the repeatability and the reproducibility revealed good to moderate reliability for tongue pressure measurement and the time of maximum tongue pressure (ICC, 0.60~0.84). However, it was poor for measurement of laryngeal movement. There were no significant differences between gender and age groups in tongue pressure, the time to maximum tongue pressure and laryngeal movement. The tongue pressure was higher in dry swallowing than in wet swallowing. Also, the time to maximum tongue pressure and overall laryngeal movement were prolonged in dry swallowing. CONCLUSION: The newly developed tongue pressure measurement system is a reliable apparatus and there are no significant age and gender differences in tongue and laryngeal movement in healthy subjects.