Dependencies and Ill-designed Parameters Within High-speed Videoendoscopy and Acoustic Signal Analysis

Schlegel, Patrick; Stingl, Michael; Kniesburges, Stefan; Bohr, Christopher; Döllinger, Michael

doi:10.1016/j.jvoice.2018.04.011

Cited by 26 publications

(26 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Numerous studies have shown significant relationships between different disorders and parameters computed from the segmentation data [33][34][35] , such as the cepstral peak prominence 11 . Typical signals derived from the glottis segmentation are the glottal area waveform (GAW) 36 , the vocal fold trajectories 37 and the phonovibrogram 38 . Parameters computed from these signals bear the promise of a higher objectivity than many of the purely subjective metrics still being employed in the clinical routine 36,[39][40][41] .…”

Section: Videoendoscopy and Glottis Segmentation Several Imaging Tecmentioning

confidence: 99%

“…Typical signals derived from the glottis segmentation are the glottal area waveform (GAW) 36 , the vocal fold trajectories 37 and the phonovibrogram 38 . Parameters computed from these signals bear the promise of a higher objectivity than many of the purely subjective metrics still being employed in the clinical routine 36,[39][40][41] . Figure 1 shows an exemplary HSV frame and the corresponding segmentation.…”

Section: Videoendoscopy and Glottis Segmentation Several Imaging Tecmentioning

confidence: 99%

See 1 more Smart Citation

BAGLS, a multihospital Benchmark for Automatic Glottis Segmentation

et al. 2020

Self Cite

View full text Add to dashboard Cite

Laryngeal videoendoscopy is one of the main tools in clinical examinations for voice disorders and voice research. Using high-speed videoendoscopy, it is possible to fully capture the vocal fold oscillations, however, processing the recordings typically involves a time-consuming segmentation of the glottal area by trained experts. Even though automatic methods have been proposed and the task is particularly suited for deep learning methods, there are no public datasets and benchmarks available to compare methods and to allow training of generalizing deep learning models. In an international collaboration of researchers from seven institutions from the EU and USa, we have created BaGLS, a large, multihospital dataset of 59,250 high-speed videoendoscopy frames with individually annotated segmentation masks. The frames are based on 640 recordings of healthy and disordered subjects that were recorded with varying technical equipment by numerous clinicians. the BaGLS dataset will allow an objective comparison of glottis segmentation methods and will enable interested researchers to train their own models and compare their methods.

show abstract

Section: Videoendoscopy and Glottis Segmentation Several Imaging Tecmentioning

confidence: 99%

Section: Videoendoscopy and Glottis Segmentation Several Imaging Tecmentioning

confidence: 99%

BAGLS, a multihospital Benchmark for Automatic Glottis Segmentation

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…To achieve this, the time-varying opening (glottis) between the oscillating vocal folds is typically extracted from subsequent images of a high-speed video sequence, further denoted as glottal area waveform (GAW) [26]. From the GAW(t) time signal quantitative measures describing the stability of the vibration pattern in respect to its vibration amplitude and cycle-periodicity as well as information about the duration of vocal fold contact time can be derived [23,[27][28][29][30]. The GAW-analysis provides first valuable information about glottal vibration characteristics but does not enable lateral comparisons of the vibration pattern of the left and right vocal fold.…”

Section: Analysis Of Vocal Fold Vibrationsmentioning

confidence: 99%

Fully automatic segmentation of glottis and vocal folds in endoscopic laryngeal high-speed videos using a deep Convolutional LSTM Network

et al. 2020

View full text Add to dashboard Cite

The objective investigation of the dynamic properties of vocal fold vibrations demands the recording and further quantitative analysis of laryngeal high-speed video (HSV). Quantification of the vocal fold vibration patterns requires as a first step the segmentation of the glottal area within each video frame from which the vibrating edges of the vocal folds are usually derived. Consequently, the outcome of any further vibration analysis depends on the quality of this initial segmentation process. In this work we propose for the first time a procedure to fully automatically segment not only the time-varying glottal area but also the vocal fold tissue directly from laryngeal high-speed video (HSV) using a deep Convolutional Neural Network (CNN) approach. Eighteen different Convolutional Neural Network (CNN) network configurations were trained and evaluated on totally 13,000 high-speed video (HSV) frames obtained from 56 healthy and 74 pathologic subjects. The segmentation quality of the best performing Convolutional Neural Network (CNN) model, which uses Long Short-Term Memory (LSTM) cells to take also the temporal context into account, was intensely investigated on 15 test video sequences comprising 100 consecutive images each. As performance measures the Dice Coefficient (DC) as well as the precisions of four anatomical landmark positions were used. Over all test data a mean Dice Coefficient (DC) of 0.85 was obtained for the glottis and 0.91 and 0.90 for the right and left vocal fold (VF) respectively. The grand average precision of the identified landmarks amounts 2.2 pixels and is in the same range as comparable manual expert segmentations which can be regarded as Gold Standard. The method proposed here requires no user interaction and overcomes the limitations of current semiautomatic or computational expensive approaches. Thus, it allows also for the analysis of long high-speed video (HSV)-sequences and holds the promise to facilitate the objective analysis of vocal fold vibrations in clinical routine. The here used dataset including the ground truth will be provided freely for all scientific groups to allow a quantitative benchmarking of segmentation approaches in future.

show abstract

“…In contrast RAP K , which is a normalized version of RAP B , does not show statistically significant changes. In a previous work it was found that the maximum reachable value of RAP K depends on the number of analyzed cycles [51], which is not the case for RAP B , if the sequence length exceeds five cycles. Hence it seems natural to assume that RAP K changes more strongly with changing sequence lengths than RAP B .…”

Section: Discussionmentioning

confidence: 89%

Influence of Analyzed Sequence Length on Parameters in Laryngeal High-Speed Videoendoscopy

et al. 2018

Self Cite

View full text Add to dashboard Cite

Laryngeal high-speed videoendoscopy (HSV) allows objective quantification of vocal fold vibratory characteristics. However, it is unknown how the analyzed sequence length affects some of the computed parameters. To examine if varying sequence lengths influence parameter calculation, 20 HSV recordings of healthy females during sustained phonation were investigated. The clinical prevalent Photron Fastcam MC2 camera with a frame rate of 4000 fps and a spatial resolution of 512 × 256 pixels was used to collect HSV data. The glottal area waveform (GAW), describing the increase and decrease of the area between the vocal folds during phonation, was extracted. Based on the GAW, 16 perturbation parameters were computed for sequences of 5, 10, 20, 50 and 100 consecutive cycles. Statistical analysis was performed using SPSS Statistics, version 21. Only three parameters (18.8%) were statistically significantly influenced by changing sequence lengths. Of these parameters, one changed until 10 cycles were reached, one until 20 cycles were reached and one, namely Amplitude Variability Index (AVI), changed between almost all groups of different sequence lengths. Moreover, visually observable, but not statistically significant, changes within parameters were observed. These changes were often most prominent between shorter sequence lengths. Hence, we suggest using a minimum sequence length of at least 20 cycles and discarding the parameter AVI.

show abstract

Dependencies and Ill-designed Parameters Within High-speed Videoendoscopy and Acoustic Signal Analysis

Cited by 26 publications

References 22 publications

BAGLS, a multihospital Benchmark for Automatic Glottis Segmentation

BAGLS, a multihospital Benchmark for Automatic Glottis Segmentation

Fully automatic segmentation of glottis and vocal folds in endoscopic laryngeal high-speed videos using a deep Convolutional LSTM Network

Influence of Analyzed Sequence Length on Parameters in Laryngeal High-Speed Videoendoscopy

Contact Info

Product

Resources

About