Characteristics of the swallowing sounds recorded in the ear, nose and on trachea

Using Machine Learning and a Combination of Respiratory Flow, Laryngeal Motion, and Swallowing Sounds to Classify Safe and Unsafe Swallowing

OBJECTIVE

The aim of this research was to develop a swallowing assessment method to help prevent aspiration pneumonia. The method uses simple sensors to monitor swallowing function during an individual's daily life.

METHODS

The key characteristics of our proposed method are as follows. First, we assess swallowing function by using respiratory flow, laryngeal motion, and swallowing sound signals recorded by simple sensors. Second, we classify whether the recorded signals correspond to healthy subjects or patients with dysphagia. Finally, we analyze the recorded signals using both a feature extraction method (linear predictive coding) and a machine learning method (support vector machine).

RESULTS

Based on our experimental results for 140 healthy subjects (54.5 32.5 years old) and 52 patients with dysphagia (75.5 20.5 years old), our proposed method could achieve 82.4% sensitivity and 86.0% specificity.

CONCLUSION

Although 20% of testing sample sets were erroneously classified, we conclude that our proposed method may facilitate screening examinations of swallowing function.

SIGNIFICANCE

In combination with the portable sensors, our proposed method is worth utilizing for noninvasive swallowing assessment.

A noninvasive swallowing measurement system using a combination of respiratory flow, swallowing sound, and laryngeal motion

The assessment of swallowing function is important for the prevention of aspiration pneumonia. We developed a new swallowing monitoring system that uses respiratory flow, swallowing sound, and laryngeal motion. We applied this device to 11 healthy volunteers and 10 patients with dysphagia. Videofluoroscopy (VF) was conducted simultaneously with swallowing monitoring using our device. We measured laryngeal rising time (LRT), the time required for the larynx to elevate to the highest position, and laryngeal activation duration (LAD), the duration between the onset of rapid laryngeal elevation and the time when the larynx returned to the lowest position. In addition, we evaluated the coordination between swallowing and breathing. We found that LAD was correlated with a VF-derived parameter, pharyngeal response duration (PRD) in healthy subjects (LAD: 959 ± 259 ms vs. PRD: 1062 ± 149 ms, r = 0.60); however, this correlation was not found in the dysphagia patients. LRT was significantly prolonged in patients (healthy subjects: 320 ± 175 ms vs. patients: 465 ± 295 ms, P < 0.001, t test). Furthermore, frequency of swallowing immediately after inspiration was significantly increased in patients. Therefore, the new device may facilitate the assessment of some aspects of swallowing dysfunction.

Dysphagia Screening: Contributions of Cervical Auscultation Signals and Modern Signal-Processing Techniques

Cervical auscultation is the recording of sounds and vibrations caused by the human body from the throat during swallowing. While traditionally done by a trained clinician with a stethoscope, much work has been put towards developing more sensitive and clinically useful methods to characterize the data obtained with this technique. The eventual goal of the field is to improve the effectiveness of screening algorithms designed to predict the risk that swallowing disorders pose to individual patients' health and safety. This paper provides an overview of these signal processing techniques and summarizes recent advances made with digital transducers in hopes of organizing the highly varied research on cervical auscultation. It investigates where on the body these transducers are placed in order to record a signal as well as the collection of analog and digital filtering techniques used to further improve the signal quality. It also presents the wide array of methods and features used to characterize these signals, ranging from simply counting the number of swallows that occur over a period of time to calculating various descriptive features in the time, frequency, and phase space domains. Finally, this paper presents the algorithms that have been used to classify this data into 'normal' and 'abnormal' categories. Both linear as well as non-linear techniques are presented in this regard.