Reliability of computerized measurements of glottal insufficiency

Rethinking glottal midline detection

A healthy voice is crucial for verbal communication and hence in daily as well as professional life. The basis for a healthy voice are the sound producing vocal folds in the larynx. A hallmark of healthy vocal fold oscillation is the symmetric motion of the left and right vocal fold. Clinically, videoendoscopy is applied to assess the symmetry of the oscillation and evaluated subjectively. High-speed videoendoscopy, an emerging method that allows quantification of the vocal fold oscillation, is more commonly employed in research due to the amount of data and the complex, semi-automatic analysis. In this study, we provide a comprehensive evaluation of methods that detect fully automatically the glottal midline. We used a biophysical model to simulate different vocal fold oscillations, extended the openly available BAGLS dataset using manual annotations, utilized both, simulations and annotated endoscopic images, to train deep neural networks at different stages of the analysis workflow, and compared these to established computer vision algorithms. We found that classical computer vision perform well on detecting the glottal midline in glottis segmentation data, but are outperformed by deep neural networks on this task. We further suggest GlottisNet, a multi-task neural architecture featuring the simultaneous prediction of both, the opening between the vocal folds and the symmetry axis, leading to a huge step forward towards clinical applicability of quantitative, deep learning-assisted laryngeal endoscopy, by fully automating segmentation and midline detection.

Laryngeal stroboscopy-Normative values for amplitude, open quotient, asymmetry and phase difference in young adults

OBJECTIVE

To provide the normative values for laryngeal stroboscopy (LS) concerning amplitude, open quotient, asymmetry and phase difference in healthy, young subjects.

STUDY DESIGN

Prospective case-control study.

SETTING

Patients treated at a single institute.

METHODS

A total of 68 healthy subjects were included in the study (35 women, 33 men), aged 18-35 years. After obtaining LS recordings, image processing was performed to attain parameters of vocal fold vibration.

RESULTS

In women, the location of the maximum vibration amplitude is approximately in the 1/3 posterior part of the glottis, while in men, the location is moved to the glottis centre. In males, the relative amplitude vibration of the vocal folds in the 1/3 anterior part of the glottis was significantly higher than in females (P = 0.029). Women showed significantly higher open quotients (OQ) at the posterior part of the glottis than the male subjects (P < 0.001) and men presented significantly higher OQ at the anterior part of the glottis than the females (P < 0.001). The average OQ values for both sexes were almost the same. Females showed significantly higher relative glottal gap area (P = 0.044). Women presented a significantly lower amplitude asymmetry than men (P = 0.002). The weighted absolute left-right phase difference reached up to 24° and remained insignificantly higher in the men than the women (P = 0.142).

CONCLUSIONS

The study provides normative values for LS in young adults for the measurement of therapy outcomes in patients with voice disorders and realisation of evidence-based medicine. The LS parametrisation is easy to perform in clinical practice.

Accuracy of measurements on digital videostroboscopic images of the vocal folds

OBJECTIVES

The reliability of objective measurements on digital laryngeal images was investigated.

METHODS

The magnitude of the error of measurement of surface areas by visually tracing the outline was determined for three different areas: the area of the lesion and, during vibration, the glottal area on maximal opening of the vocal folds and the glottal area on maximal closing of the vocal folds.

RESULTS

The errors in these areas were 10% to 30%. The results suggest that the error is mainly due to the uncertainty of the real outline of areas with a vague boundary. Correction for differences in magnification between two images (posttherapy and pretherapy) is of importance in about 25% to 65% of cases, depending on the area measured.

CONCLUSIONS

Only when the magnification ratio is small (less than about +/- 10% from 1.0) may a correction not be necessary.

Cross-linked hyaluronan versus collagen for injection treatment of glottal insufficiency: 2-year follow-up

OBJECTIVES

To evaluate the long-term (24 months) clinical performance (vocal fold function) and safety of hylan B gel as compared with bovine cross-linked collagen in the treatment of patients with glottal insufficiency.

MATERIAL AND METHODS

In a prospective trial, 70 patients with glottal insufficiency due to unilateral vocal fold paresis (n =35) or atrophy (n =35) were randomized to received either hylan B gel (n =47) or collagen (n =23) injections into 1 vocal fold. Forty-two of the patients were examined 24 months after treatment. Evaluations were made based on patients' subjective ratings, digitized videostroboscopic measurements, maximum phonation time and phonation quotient.

RESULTS

The patients' self-ratings were significantly improved in both the hylan B gel and collagen groups. Videostroboscopic measurements of glottal closure were significantly improved for both groups. The hylan B gel group showed a trend towards less resorption at the injected vocal fold edge in comparison with the collagen group (p =0.05). No serious adverse events were observed. Twenty-eight patients dropped out of the study after 12 months: 18 had been re-injected or operated on with medialization laryngoplasty due to insufficient voice and 10 had either died of causes unrelated to the study or refused to attend follow-up.

CONCLUSIONS

No long-term side-effects were found for either the hylan B gel or collagen groups after injection treatment. Both treatments resulted in significantly improved voice as rated by the patients and significantly improved glottal closure. Some resorption was noted for both substances, and approximately 25%, of the patients chose re-treatment 2 years after the initial treatment.

Vibratory capacity and voice outcome in patients with scarred vocal folds treated with collagen injections--case studies

This prospective case study evaluates the effect of collagen injections on vibratory capacity and voice outcome in four patients with vocal fold scarring. The minimum follow-up time was 6 months. We studied the effect of the injections on the vibratory capacity in a multidimensional way, using computerized image analysis of videostroboscopy, acoustic and perceptual analysis of voice recordings, as well as patients' self-evaluations. Subjective voice ratings from the patients showed improvement in two patients, while two patients had no change in voice function. None of the patients experienced any side effects from the treatment. The documented acoustic and perceptual changes were marginal, but the videostroboscopic findings suggest some improvement in vibratory capacity.

Cross-linked hyaluronan used as augmentation substance for treatment of glottal insufficiency: safety aspects and vocal fold function

OBJECTIVE

To examine safety aspects and vocal fold function after vocal fold augmentation with a cross-linked hyaluronan derivative (hylan B gel) as compared with bovine collagen. STUDY DESIGN; A prospective, randomized trial.

METHODS

Eighty-three patients with glottal insufficiency were treated with injection augmentation with hylan B gel and bovine collagen and were examined at 1, 6, and 12 months after treatment. Seventy patients with unilateral vocal fold paresis (n = 35) or atrophy (n = 35) were randomly assigned to receive either hylan B gel (n = 47) or collagen (n = 23) injections into one vocal fold. Thirteen patients with glottal insufficiency caused by scar defects or paresis resulting from malignant disease were included in a nonrandomized group and were treated only with hylan B gel. Evaluations were made from patients' subjective ratings (visual analogue scales), digitized videostroboscopic measurements, phonetograms, maximum phonation time, and phonation quotients.

RESULTS

Twelve months after injections, the patients' self-ratings were significantly improved for both the hylan B gel and the collagen groups. In addition, the videostroboscopic measurements showed significantly improved glottal closure for both groups. However, for the hylan B gel group, vibration amplitude and glottal area variations were preserved, and this group showed significantly less resorption at the injected vocal fold edge. Furthermore, maximum phonation time had increased significantly for the hylan B gel patients (collagen, nonsignificant). No serious adverse events were observed; three patients injected with hylan B gel had temporary inflammation at the injection site, which resolved without sequelae.

CONCLUSIONS

The results showed that both hylan B gel and collagen can be safely used for injection treatment of glottal insufficiency. Both treatments resulted in significantly improved voice as rated by the patients. However, the patients treated with hylan B gel showed better vocal fold status and longer maximum phonation time at 12 months after treatment as compared with patients treated with collagen.

Effects of voice therapy as objectively evaluated by digitized laryngeal stroboscopic imaging

Objective measurements derived from digitized laryngeal stroboscopic images were used to demonstrate changes in vocal fold vibration and in the size of benign lesions after 3 months of voice therapy. Forty chronically dysphonic patients were studied. By means of a rigid stroboscope, pretreatment and posttreatment recordings were made of the vocal folds at rest and under stroboscopic light during phonation. From each recording, images of the positions at rest and during vibration at maximal opening and at maximal closure were digitized. The surface areas of any lesions and of the glottal gap were independently measured in the digitized images by 2 experienced laryngologists. Referential distances were determined in order to compensate for discrepancies in magnification in the various recordings. After 3 months of voice therapy, significant improvement in lesion size and degree of maximal closure during vibration could be demonstrated in about 50% of the patients. The degree of maximal opening did not prove to be a significant parameter.