Machine Learning for Automatic Detection of Velopharyngeal Dysfunction: A Preliminary Report.

BACKGROUND: Even after palatoplasty, the incidence of velopharyngeal dysfunction (VPD) can reach 30%; however, these estimates arise from high-income countries (HICs) where speech-language pathologists (SLP) are part of standardized cleft teams. The VPD burden in low- and middle-income countries (LMICs) is unknown. This study aims to develop a machine-learning model that can detect the presence of VPD using audio samples alone. METHODS: Case and control audio samples were obtained from institutional and publicly available sources. A machine-learning model was built using Python software. RESULTS: The initial 110 audio samples used to test and train the model were retested after format conversion and file deidentification. Each sample was tested 5 times yielding a precision of 100%. Sensitivity was 92.73% (95% CI: 82.41%-97.98%) and specificity was 98.18% (95% CI: 90.28%-99.95%). One hundred thirteen prospective samples, which had not yet interacted with the model, were then tested. Precision was again 100% with a sensitivity of 88.89% (95% CI: 78.44%-95.41%) and a specificity of 66% (95% CI: 51.23%-78.79%). DISCUSSION: VPD affects nearly 100% of patients with unrepaired overt soft palatal clefts and up to 30% of patients who have undergone palatoplasty. VPD can render patients unintelligible, thereby accruing significant psychosocial morbidity. The true burden of VPD in LMICs is unknown, and likely exceeds estimates from HICs. The ability to access a phone-based screening machine-learning model could expand access to diagnostic, and potentially therapeutic modalities for an innumerable amount of patients worldwide who suffer from VPD.

Optimizing Botox regimens in patients with adductor spasmodic dysphonia and essential tremor of voice: A 31-year experience.

Objective: The purpose of this study was to quantitatively compare the effectiveness of unilateral and bilateral botulinum toxin A (BTX-A) injections for mitigating undesirable weak/breathy voice quality and dysphagia for patients with adductor spasmodic dysphonia and/or essential tremor of voice (ETV). Methods: Data were collected from the medical records of 319 patients, yielding three treatment cohorts: patients who received an equal dose bilateral injection regimen (BL=) throughout their course of treatment at VUMC, patients who switched to a unilateral injection regimen (UL), and patients who switched to an unequal dose bilateral injection regimen (BL≠). Changes in length of improvement, duration of weak/breathy voice, and dysphagia severity were compared. Results: The BL = treatment group reported the longest duration of improved voice. Shorter periods of improved voice were reported at baseline by patients who later switched to UL or BL ≠ injection regimens. Patients receiving UL injections reported significantly reduced weak/breathy voice and dysphagia. Patients receiving BL ≠ injections reported increased length of improved voice; however, dysphagia symptoms increased. Ninety-two percent of patients with ETV switched to a UL regimen, with 61% of patients transitioning within the first three injections. Conclusions: Patients with pronounced dysphagia and extended periods of weak/breathy voice may benefit from a UL injection approach to mitigate side effects from BTX-A without sacrificing improved voice outcomes. For patients seeking to extend their length of improved voice, a BL ≠ injection regimen may be effective provided the adverse side effects from BTX-A are minimal. Patients with ETV may benefit from a UL injection approach at the outset of their course of treatment with BTX-A. Level of evidence: III.

Continuous Rate Infusion of Ketamine Hydrochloride and Dexmedetomidine for Maintenance of Anesthesia during Laryngotracheal Surgery in New Zealand White Rabbits (Oryctolagus cuniculus).

New Zealand white rabbits (Oryctolagus cuniculus) are an established in vivo model for the study of structural and functional consequences of vocal-fold vibration. Research design requires invasive laryngotracheal procedures, and the presence of laryngospasms or pain responses (or both) hinder phonation-related data collection. Published anesthesia regimens report respiratory depression and muscle tone changes and have been unsuccessful in mitigating autonomic laryngeal responses in our protocol. Infusion of ketamine hydrochloride and dexmedetomidine hydrochloride in pediatric medicine provides effective analgesia and sedation for laryngotracheal procedures including intubation and bronchoscopy; however, data evaluating the use of ketamine-dexmedetomidine infusion in rabbits are unavailable. This study reports a new infusion regimen, which was used in 58 male New Zealand white rabbits that underwent a nonsurvival laryngotracheal procedure to induce phonotraumatic vocal-fold injury. Animals were sedated by using ketamine hydrochloride (20 mg/kg IM) and dexmedetomidine (0.125 mg/kg IM). Maintenance anesthesia was provided by using continuous rate intravenous infusion of ketamine hydrochloride (343 µg/kg/min) and dexmedetomidine (1.60 µg/kg/min). A stable plane of anesthesia with no autonomic laryngeal response (laryngospasm) was achieved in 32 of the 58 rabbits (55%). Laryngospasms occurred in 25 of 58 animals (43%) and were controlled in 20 cases (80%) by providing 0.33 mL 2% topical lidocaine, incremental increase in infusion rate, or both. Continuous rate infusion of ketamine hydrochloride-dexmedetomidine with prophylactic topical lidocaine provides a predictable and adjustable surgical plane of anesthesia, with minimal confounding respiratory and autonomic laryngeal responses, during extended-duration laryngotracheal surgery in rabbits. This regimen should be considered as an alternative to injection maintenance for prolonged, invasive procedures.

Unilateral and Bilateral Laryngeal Pacing for Bilateral Vocal Fold Paralysis.

PURPOSE OF REVIEW: Present the state-of-the-art overview of laryngeal pacing for treatment of bilateral vocal fold paralysis. A minimally invasive unilateral pacing system and a fully implantable bilateral pacing system are currently in clinical trials. The relative advantages and disadvantages of each are discussed. RECENT FINDINGS: Research in functional electrical stimulation for the reanimation of the posterior cricoarytenoid muscle has successfully translated from animal models to human clinical trials for unilateral pacing and bilateral pacing. Current findings suggest unilateral pacing in humans significantly improves ventilation but only marginally better than cordotomy. Bilateral pacing in canines increases glottal opening greater than 2-fold over unilateral pacing and restores exercise tolerance to normal. SUMMARY: Unilateral pacing can be considered a breathing assist device and may not be appropriate for active individuals. Bilateral pacing may be preferable for patients who wish to engage in strenuous exercise. Minimally invasive systems may be ideal for patients who prefer less invasive implantation and are not concerned with cosmesis. Fully implantable pacing systems offer greater electrode redundancy and stability, resulting in a system that is robust against electrode migration or damage.

Efficacy of Videostroboscopy and High-Speed Videoendoscopy to Obtain Functional Outcomes From Perioperative Ratings in Patients With Vocal Fold Mass Lesions.

OBJECTIVES: A major limitation of comparing the efficacy of videostroboscopy (VS) and high-speed videoendoscopy (HSV) is the lack of an objective reference by which to compare the functional assessment ratings of the two techniques. For patients with vocal fold mass lesions, intraoperative measures of lesion size and depth may serve as this objective reference. This study compared the relationships between the pre- to postoperative change in VS and HSV visual-perceptual ratings to intraoperative measures of lesion size and depth. DESIGN: Prospective visual-perceptual study with intraoperative measures of lesion size and depth. METHODS: VS and HSV samples were obtained preoperatively and postoperatively from 28 patients with vocal fold lesions and from 17 vocally healthy controls. Two experienced clinicians rated amplitude, mucosal wave, vertical phase difference, left-right phase asymmetry, and vocal fold edge on a visual-analog scale using both imaging techniques. The change in perioperative ratings from VS and HSV was compared between groups and correlated to intraoperative measures of lesion size and depth. RESULTS: HSV was as reliable as VS for ratings of amplitude and edge, and substantially more reliable for ratings of mucosal wave and left-right phase asymmetry. Both VS and HSV had mild-moderate correlations between change in perioperative ratings and intraoperative measures of lesion area. Change in function could be obtained in more patients and for more parameters using HSV than VS. Group differences were noted for postoperative ratings of amplitude and edge; however, these differences were within one level of the visual-perceptual rating scale. The presence of asynchronicity in VS recordings renders vibratory features either uninterpretable or potentially distorted and thus should not be rated. CONCLUSIONS: Amplitude and edge are robust vibratory measures for perioperative functional assessment, regardless of imaging modality. HSV is indicated for evaluation of subepithelial lesions or if asynchronicity is present in the VS image sequence.

Decoding phonation with artificial intelligence (DeP AI): Proof of concept.

OBJECTIVE: Acoustic analysis of voice has the potential to expedite detection and diagnosis of voice disorders. Applying an image-based, neural-network approach to analyzing the acoustic signal may be an effective means for detecting and differentially diagnosing voice disorders. The purpose of this study is to provide a proof-of-concept that embedded data within human phonation can be accurately and efficiently decoded with deep learning neural network analysis to differentiate between normal and disordered voices. METHODS: Acoustic recordings from 10 vocally-healthy speakers, as well as 70 patients with one of seven voice disorders (n = 10 per diagnosis), were acquired from a clinical database. Acoustic signals were converted into spectrograms and used to train a convolutional neural network developed with the Keras library. The network architecture was trained separately for each of the seven diagnostic categories. Binary classification tasks (ie, to classify normal vs. disordered) were performed for each of the seven diagnostic categories. All models were validated using the 10-fold cross-validation technique. RESULTS: Binary classification averaged accuracies ranged from 58% to 90%. Models were most accurate in their classification of adductor spasmodic dysphonia, unilateral vocal fold paralysis, vocal fold polyp, polypoid corditis, and recurrent respiratory papillomatosis. Despite a small sample size, these findings are consistent with previously published data utilizing deep neural networks for classification of voice disorders. CONCLUSION: Promising preliminary results support further study of deep neural networks for clinical detection and diagnosis of human voice disorders. Current models should be optimized with a larger sample size. LEVELS OF EVIDENCE: Level III.

Vibratory function and healing outcomes after small intestinal submucosa biomaterial implantation for chronic vocal fold scar.

OBJECTIVES/HYPOTHESIS: Vocal fold scar is a major cause of dysphonia, and optimal treatments do not currently exist. Small intestinal submucosa (SIS) is a biomaterial developed for the treatment of a variety of pathologies. The purpose of this study was to investigate the effects of SIS implantation on tissue remodeling in scarred vocal folds using routine staining, immunohistochemistry, and high-speed videoendoscopy (HSV). STUDY DESIGN: Prospective, blinded group analysis. METHODS: Thirteen New Zealand White rabbits underwent a vocal fold scarring procedure followed by microflap elevation with or without SIS implantation. Seven months later, they underwent a phonation procedure with HSV and laryngeal harvest. Alcian blue and elastica van Gieson staining and immunohistochemistry for collagen types I and III were used to evaluate histological healing outcomes. Dynamic functional remodeling of the scarred vocal fold in the presence of SIS implants was evaluated using HSV imaging to capture restoration of vibratory amplitude, amplitude ratio, and left-right phase symmetry. RESULTS: Density of collagen I was significantly decreased in SIS versus microflap-treated vocal folds. No differences were found between groups for hyaluronic acid, elastin, or collagen type III. Organization of elastin in the subepithelial region appeared to affect amplitude of vibration and the shape of the vocal fold edge. CONCLUSIONS: SIS implantation into chronic scar reduced the density of collagen I deposits. There was no evidence of a negative impact or complication from SIS implantation. Regardless of treatment type, organization of elastin in the subepithelial region may be important to vibratory outcomes. LEVEL OF EVIDENCE: NA. Laryngoscope, 128:901-908, 2018.

Comparison of Videostroboscopy to Stroboscopy Derived From High-Speed Videoendoscopy for Evaluating Patients With Vocal Fold Mass Lesions.

Purpose: Videostroboscopy (VS) uses an indirect physiological signal to predict the phase of the vocal fold vibratory cycle for sampling. Simulated stroboscopy (SS) extracts the phase of the glottal cycle directly from the changing glottal area in the high-speed videoendoscopy (HSV) image sequence. The purpose of this study is to determine the reliability of SS relative to VS for clinical assessment of vocal fold vibratory function in patients with mass lesions. Methods: VS and SS recordings were obtained from 28 patients with vocal fold mass lesions before and after phonomicrosurgery and 17 controls who were vocally healthy. Two clinicians rated clinically relevant vocal fold vibratory features using both imaging techniques, indicated their internal level of confidence in the accuracy of their ratings, and provided reasons for low or no confidence. Results: SS had fewer asynchronous image sequences than VS. Vibratory outcomes were able to be computed for more patients using SS. In addition, raters demonstrated better interrater reliability and reported equal or higher levels of confidence using SS than VS. Conclusion: Stroboscopic techniques on the basis of extracting the phase directly from the HSV image sequence are more reliable than acoustic-based VS. Findings suggest that SS derived from high-speed videoendoscopy is a promising improvement over current VS systems.