Clinical evaluation of a machine learning-based dysphagia risk prediction tool.

PURPOSE: The rise of digitization promotes the development of screening and decision support tools. We sought to validate the results from a machine learning based dysphagia risk prediction tool with clinical evaluation. METHODS: 149 inpatients in the ENT department were evaluated in real time by the risk prediction tool, as well as clinically over a 3-week period. Patients were classified by both as patients at risk/no risk. RESULTS: The AUROC, reflecting the discrimination capability of the algorithm, was 0.97. The accuracy achieved 92.6% given an excellent specificity as well as sensitivity of 98% and 82.4% resp. Higher age, as well as male sex and the diagnosis of oropharyngeal malignancies were found more often in patients at risk of dysphagia. CONCLUSION: The proposed dysphagia risk prediction tool proved to have an outstanding performance in discriminating risk from no risk patients in a prospective clinical setting. It is likely to be particularly useful in settings where there is a lower incidence of patients with dysphagia and less awareness among staff.

Effects of Finding the Speech-Language Pathologist Likeable on Postlaryngectomy Speech Intelligibility Outcomes.

INTRODUCTION: Speech-language pathologists (SLPs) work with patients after total laryngectomy (TL) to regain verbal communication. The influence of the quality of the therapeutic relationship on the success of TL voice rehabilitation in terms of speech intelligibility is not known. Finding each other likeable is an important factor in establishing and maintaining interpersonal relationships in everyday life. The fit of therapist and client is relevant to the therapeutic relationship. The purpose of this study therefore was to assess the association between the degree of SLPs' likeability ratings and postlaryngectomy speech intelligibility. METHODS: In a multicentre prospective cohort study, participants rated their SLPs' likeability after finishing TL rehabilitation. Speech intelligibility was measured objectively with the Post-Laryngectomy Telephone Intelligibility Test and subjectively with the Questionnaire for Adjustment after Laryngectomy. The association of SLPs' likeability with speech intelligibility was analysed using hierarchical logistic regression, expressed with odds ratios (OR) with corresponding 95% confidence intervals (CI). RESULTS: Altogether 124 patients from 13 institutions participated. The degree of finding the SLP likeable was not significantly associated with objective speech intelligibility (OR 1.30; 95% CI 0.78-2.18; p = 0.32) or subjective speech intelligibility (OR 1.01; 95% CI 0.60-1.72; p = 0.96) after controlling for age, sex and education factors. DISCUSSION/CONCLUSION: In this patient cohort, there was no evidence for an association between ratings of SLPs' likeability and speech intelligibility outcomes after rehabilitation. Future studies could consider the use of alternative instruments for measuring likeability.

Functional Electrical Stimulation for Presbyphonia: A Prospective Randomized Trial.

OBJECTIVES: The aim of this prospective study was to examine the effects of transcutaneous functional electrical stimulation (FES) in a group of elderly women with presbyphonia. STUDY DESIGN: Prospective randomized study. METHODS: Fourteen participants were enrolled prospectively and attributed randomly to two different treatment groups, where one group (n = 7) received 8 weeks of training (5 days a week), whereas the other group (n = 7) received 4 weeks of ineffective stimulation, followed by 4 weeks of effective training. Stimulation protocols were established during baseline examination and confirmed with endoscopy to ensure a glottal reaction. Numerous acoustical, vocal, patient-centered, and respiratory parameters were obtained at several time points. RESULTS: Neither 4 weeks nor 8 weeks of functional electrical transcutaneous stimulation led to changes of vocal, acoustical, or respiratory parameters, apart from patient-centered items (Voice Handicap Index 12, Voice-Related Quality of Life), which improved over time. However, there were no differences between the two arms for both items. CONCLUSIONS: Transcutaneous FES over 4 weeks and 8 weeks did not lead to significantly improved objective voice and acoustical parameters, which could be caused by the fact that the muscles of interest cannot be targeted specifically enough. However, we found a significant improvement of subjective voice perception and voice-related quality of life in both groups. We explain this finding with an observer-expectancy effect secondary to the very time-consuming and elaborate study procedures. LEVEL OF EVIDENCE: 1b Laryngoscope, 130:E662-E666, 2020.

Exploring stimulation patterns for electrical stimulation of the larynx using surface electrodes.

PURPOSE: Functional electrical stimulation (FES) is considered an upcoming treatment modality for a number of laryngeal diseases. However, sound data are scarce when it comes to surface FES to treat voice disorders. Aim of the present study was to identify and differentiate suitable surface FES patterns to activate internal laryngeal muscles. METHODS: Non-invasive FES was performed in a cohort of 17 elderly woman. Our user-customized electrical stimulation setup allowed us to deliver ten different stimulation patterns (rectangular and sawtooth shaped) with variation of frequency and amplitude. Stimulation outcome, i.e., vocal fold (VF) reaction, was continuously verified by transnasal endoscopy. RESULTS: Responses to FES using ten different stimulation patterns varied inter-individually. None of the stimulation parameter sets could elicit a VF reaction in all participants. CONCLUSION: Based on our findings we conclude that individual fitting is necessary when defining surface stimulation parameters. To overcome limitations of previous studies, devices with freely programmable patterns are required as shown here. Endoscopic control of VF reaction is absolutely essential to ensure effectiveness of the delivered patterns.

Segmenting and tracking cell instances with cosine embeddings and recurrent hourglass networks.

Differently to semantic segmentation, instance segmentation assigns unique labels to each individual instance of the same object class. In this work, we propose a novel recurrent fully convolutional network architecture for tracking such instance segmentations over time, which is highly relevant, e.g., in biomedical applications involving cell growth and migration. Our network architecture incorporates convolutional gated recurrent units (ConvGRU) into a stacked hourglass network to utilize temporal information, e.g., from microscopy videos. Moreover, we train our network with a novel embedding loss based on cosine similarities, such that the network predicts unique embeddings for every instance throughout videos, even in the presence of dynamic structural changes due to mitosis of cells. To create the final tracked instance segmentations, the pixel-wise embeddings are clustered among subsequent video frames by using the mean shift algorithm. After showing the performance of the instance segmentation on a static in-house dataset of muscle fibers from H&E-stained microscopy images, we also evaluate our proposed recurrent stacked hourglass network regarding instance segmentation and tracking performance on six datasets from the ISBI celltracking challenge, where it delivers state-of-the-art results.