ARTIFICIAL INTELLIGENCE AND DYSPHAGIA: NOVEL SOLUTIONS TO OLD PROBLEMS

Post-Radiotherapy Dysphagia in Head and Neck Cancer: Current Management by Speech-Language Pathologists

OPINION STATEMENT

Dysphagia, difficulty in eating and drinking, remains the most common side effect of radiotherapy treatment for head and neck cancer (HNC) with devastating consequences for function and quality of life (QOL). Over the past decade, 5-year survival has improved due to multiple factors including treatment advances, reduction in smoking, introduction of the human papillomavirus (HPV) vaccine and more favourable prognosis of HPV-related cancers. Increased prevalence of HPV-positive disease, which tends to affect younger individuals, has led to an elevated number of people living for longer with the sequelae of cancer and its treatment. Symptoms are compounded by late effects of radiotherapy which may lead to worsening of dysphagia for some long-term survivors or new-onset dysphagia for others. Speech-language pathology (SLP) input remains core to the assessment and management of dysphagia following HNC treatment. In this article, we present current SLP management of dysphagia post-radiotherapy. We discuss conventional treatment approaches, the emergence of therapy adjuncts and current service delivery models. The impact of adherence on therapy outcomes is highlighted. Despite treatment advancements, patients continue to present with dysphagia which is resistant to existing intervention approaches. There is wide variation in treatment programmes, with a paucity of evidence to support optimal type, timing and intensity of treatment. We discuss the need for further research, including exploration of the impact of radiotherapy on the central nervous system (CNS), the link between sarcopenia and radiotherapy-induced dysphagia and the benefits of visual biofeedback in rehabilitation.

PECI-Net: Bolus segmentation from video fluoroscopic swallowing study images using preprocessing ensemble and cascaded inference

Bolus segmentation is crucial for the automated detection of swallowing disorders in videofluoroscopic swallowing studies (VFSS). However, it is difficult for the model to accurately segment a bolus region in a VFSS image because VFSS images are translucent, have low contrast and unclear region boundaries, and lack color information. To overcome these challenges, we propose PECI-Net, a network architecture for VFSS image analysis that combines two novel techniques: the preprocessing ensemble network (PEN) and the cascaded inference network (CIN). PEN enhances the sharpness and contrast of the VFSS image by combining multiple preprocessing algorithms in a learnable way. CIN reduces ambiguity in bolus segmentation by using context from other regions through cascaded inference. Moreover, CIN prevents undesirable side effects from unreliably segmented regions by referring to the context in an asymmetric way. In experiments, PECI-Net exhibited higher performance than four recently developed baseline models, outperforming TernausNet, the best among the baseline models, by 4.54% and the widely used UNet by 10.83%. The results of the ablation studies confirm that CIN and PEN are effective in improving bolus segmentation performance.

Dysphagia Advances in Head and Neck Cancer

Purpose of Review

This review summarises the current literature regarding head and neck cancer-associated dysphagia. Up-to-date evidence for dysphagia outcome measurement for this population is provided, in addition to recent innovations that aim to prevent, reduce or remediate the common and debilitating side effects of treatment.

Recent Findings

Both patient-reported outcomes and clinical measures are necessary to capture the multi-dimensional nature of swallowing. A minimally important difference in scores has been calculated for some of these measures, to aid interpretation and powering of clinical trials. The number of dysphagia-related trials has increased, predominantly investigating optimal treatment for oropharyngeal HPV-positive disease, and speech and language pathology interventions using an impairment-based approach.

Summary

Although substantial progress has been made, further work is necessary to establish a consensus over outcome measures. Modifying treatments may improve outcomes. Several trials are underway to establish the effectiveness of speech and language pathology dysphagia interventions.

Accuracy of Acoustic Evaluation of Swallowing as a Diagnostic Method of Dysphagia in Individuals Affected by Stroke: Preliminary Analysis

After a stroke, more than half of the patients have some kind of disability, and dysphagia is frequently found. Cervical auscultation by Doppler sonar is an innovative technique with gain of credibility in the clinical evaluation of swallowing. To verify the diagnostic accuracy of Doppler sonar along with the DeglutiSom® software as an auxiliary method in the evaluation of oropharyngeal dysphagia in patients after stroke. The research is a cross-sectional, uncontrolled, blind, quantitative study with systematic random sampling. Patients from inpatient and outpatient units of a reference hospital with a stroke care unit were concomitantly submitted to both Doppler sonar and Fiberoptic Endoscopic Evaluation of Swallowing (FEES®). Seventy-three audio files collected from 26 patients through Doppler sonar were analyzed using DeglutiSom® software and confronted with the FEES® report, regarding three food consistencies offered to them during the exam. The study showed that the Doppler sonar correctly identified, among all the analyzed files, those that actually presented tracheal aspiration as well as it effectively identified patients who did not aspirate. The Youden index of 0.91 corroborates this information, showing a promising accuracy in detecting tracheal aspiration in the studied sample. The study evaluates the diagnostic accuracy of Doppler sonar, showing that it can be used as a valuable tool in the diagnosis of tracheal aspiration in patients after stroke. It is important to emphasize that the identification of residue by this method requires further studies. Also, larger sample size and more than one blind evaluator should be considered in future researches to increase the reliability of the proposed method.