A pilot observation using ultrasonography and vowel articulation to investigate the influence of suspected obstructive sleep apnea on upper airway.

Failure to employ suitable measures before administering full anesthesia to patients with obstructive sleep apnea (OSA) who are undergoing surgery may lead to developing complications after surgery. Therefore, it is very important to screen OSA before performing a surgery, which is currently done by subjective questionnaires such as STOP-Bang, Berlin scores. These questionnaires have 10-36% specificity in detecting sleep apnea, along with no information given on anatomy of upper airway, which is important for intubation. To address these challenges, we performed a pilot study to understand the utility of ultrasonography and vowel articulation in screening OSA. Our objective was to investigate the influence of OSA risk factors in vowel articulation through ultrasonography and acoustic features analysis. To accomplish this, we recruited 18 individuals with no risk of OSA and 13 individuals with high risk of OSA and asked them to utter vowels, such as /a/ (as in "Sah"), /e/ (as in "See"). An expert ultra-sonographer measured the parasagittal anterior-posterior (PAP) and transverse diameter of the upper airway. From the recorded vowel sounds, we extracted 106 features, including power, pitch, formant, and Mel frequency cepstral coefficients (MFCC). We analyzed the variation of the PAP diameters and vowel features from "See: /i/" to "Sah /a/" between control and OSA groups by two-way repeated measures ANOVA. We found that, there was a variation of upper airway diameter from "See" to "Sah" was significantly smaller in OSA group than control group (OSA: ∆12.8 ± 5.3 mm vs. control: ∆22.5 ± 3.9 mm OSA, p < 0.01). Moreover, we found several vowel features showed the exact same or opposite trend as PAP diameter variation, which led us to build a machine learning model to estimate PAP diameter from vowel features. We found a correlation coefficient of 0.75 between the estimated and measured PAP diameter after applying four estimation models and combining their output with a random forest model, which showed the feasibility of using acoustic features of vowel sounds to monitor upper airway diameter. Overall, this study has proven the concept that ultrasonography and vowel sounds analysis may be useful as an easily accessible imaging tool of upper airway.

The Effect of Simulated Obstructive Apneas on Mechanical Characteristics of Lower Airways in Individuals with Asthma.

Increased negative intrathoracic pressure that occurs during pharyngeal obstruction can increase thoracic fluid volume that may contribute to lower airway narrowing in individuals with obstructive sleep apnea (OSA) and asthma. Our previous study showed that fluid accumulation in the thorax induced by simulated OSA can increase total respiratory resistance. However, the effect of fluid shift on lower airway narrowing has not been investigated. To examine the effect of fluid accumulation in the thorax on the resistance of the lower airway. Non-asthma participants and individuals with (un)controlled asthma were recruited and underwent a single-day experiment. A catheter with six pressure sensors was inserted through the nose to continuously measure pressure at different sites of the airway, while a pneumotachograph was attached to a mouthpiece to record airflow. To simulate obstructive apneas, participants performed 25 Mueller maneuvers (MMs) while lying supine. Using the recordings of pressure sensor and airflow, total respiratory (RT), lower respiratory components (RL), and upper airway (RUA) resistances were calculated before and after MMs. Generalized estimation equation method was used to find the predictors of RL among variables including age, sex, body mass index, and the effect of MMs and asthma. Eighteen participants were included. Performing MMs significantly increased RT (2.23 ± 2.08 cmH2O/L/s, p = 0.003) and RL (1.52 ± 2.00 cmH2O/L/s, p = 0.023) in participants with asthma, while only RL was increased in non-asthma group (1.96 ± 1.73 cmH2O/L/s, p = 0.039). We found the model with age, and the effect of MMs and asthma severity generated the highest correlation (R2 = 0.69, p < 0.001). We provide evidence that fluid accumulation in the thorax caused by excessive intrathoracic pressure increases RL in both non-asthma and asthma groups. The changes in RL were related to age, having asthma and the effect of simulated OSA. This can explain the interrelationship between OSA and asthma.

Sleep Apnea Detection by Tracheal Motion and Sound, and Oximetry via Application of Deep Neural Networks.

Purpose: Sleep apnea (SA) is highly prevalent, but under diagnosed due to inaccessibility of sleep testing. To address this issue, portable devices for home sleep testing have been developed to provide convenience with reasonable accuracy in diagnosing SA. The objective of this study was to test the validity a novel portable sleep apnea testing device, BresoDX1, in SA diagnosis, via recording of trachea-sternal motion, tracheal sound and oximetry. Patients and Methods: Adults with a suspected sleep disorder were recruited to undergo in-laboratory polysomnography (PSG) and a simultaneous BresoDX1 recording. Data from BresoDX1 were collected and features related to breathing sounds, body motions and oximetry were extracted. A deep neural network (DNN) model was trained with 61-second epochs of the extracted features to detect apneas and hypopneas from which an apnea-hypopnea index (AHI) was calculated. The AHI estimated by BresoDX1 (AHIbreso) was compared to the AHI determined from PSG (AHIPSG) and the sensitivity and specificity of SA diagnosis were assessed at an AHIPSG ≥ 15. Results: Two-hundred thirty-three participants (mean ± SD) 50 ± 16 years of age, with BMI of 29.8 ± 6.6 and AHI of 19.5 ± 22.7, were included. There was a strong relationship between AHIbreso and AHIPSG (r = 0.91, p < 0.001). SA detection for an AHIPSG ≥ 15 was highly sensitive (90.0%) and specific (85.9%). Conclusion: We conclude that the DNN model we developed via recording and analyses of trachea-sternal motion and sound along with oximetry provides an accurate estimate of the AHIPSG with high sensitivity and specificity. Therefore, BresoDX1 is a simple, convenient and accurate portable SA monitoring device that could be employed for home SA testing in the future.

Respiratory Motion and Airflow Estimation During Sleep Using Tracheal Movement and Sound.

Purpose: Due to lack of access and high cost of polysomnography, portable sleep apnea testing has been developed to diagnose sleep apnea. Despite being less expensive, and having fewer sensors and reasonable accuracy in identifying sleep apnea, such devices can be less accurate than polysomnography in detecting apneas/hypopneas. To increase the accuracy of apnea/hypopnea detection, an accurate airflow estimation is required. However, current airflow measurement techniques employed in portable devices are inconvenient and subject to displacement during sleep. In this study, algorithms were developed to estimate respiratory motion and airflow using tracheo-sternal motion and tracheal sounds. Patients and Methods: Adults referred for polysomnography were included. Simultaneous to polysomnography, a patch device with an embedded 3-dimensional accelerometer and microphone was affixed to the suprasternal notch to record tracheo-sternal motion and tracheal sounds, respectively. Tracheo-sternal motion was used to train two mathematical models for estimating changes in respiratory motion and airflow compared to simultaneously measured thoracoabdominal motion and nasal pressure from polysomnography. The amplitude of the estimated airflow was then adjusted by the tracheal sound envelope in segments with unstable breathing. Results: Two hundred and fifty-two subjects participated in this study. Overall, the algorithms provided highly accurate estimates of changes in respiratory motion and airflow with mean square errors (MSE) of 3.58 ± 0.82% and 2.82 ± 0.71%, respectively, compared to polysomnographic signals. The estimated motion and airflow from the patch signals detected apneas and hypopneas scored on polysomnography in 63.9% and 88.3% of cases, respectively. Conclusion: This study presents algorithms to accurately estimate changes in respiratory motion and airflow, which provides the ability to detect respiratory events during sleep. Our study suggests that such a simple and convenient method could be used for portable monitoring to detect sleep apnea. Further studies will be required to test this possibility.

Association of Obstructive Apnea with Thoracic Fluid Shift and Small Airways Narrowing in Asthma During Sleep.

Rationale: Obstructive sleep apnea (OSA) is highly prevalent among patients with asthma, suggesting a pathophysiological link between the two, but a mechanism for this has not been identified. Hypothesis: Among patients with asthma, those with OSA will have greater overnight increases in thoracic fluid volume and small airways narrowing than those without OSA. Methods: We enrolled 19 participants with asthma: 9 with OSA (apnea-hypopnea index (AHI) ≥10) and 10 without OSA (AHI <10). All participants underwent overnight polysomnography. Before and after sleep, thoracic fluid volume was measured by bioelectrical impedance and small airways narrowing was primarily assessed by respiratory system reactance at 5Hz using oscillometry. Results: Patients with asthma and OSA (OSA group) had a greater overnight increase in thoracic fluid volume by 120.5 mL than patients without OSA (non-OSA group) (164.4 ± 44.0 vs 43.9 ± 47.3 mL, p=0.006). Compared to the non-OSA group, the OSA group had greater overnight decrease in reactance at 5Hz (-1.08 ± 0.75 vs 0.21 ± 0.27 cmH2O/L/s, p=0.02), and overnight increase in reactance area (14.81 ± 11.09 vs -1.20 ± 2.46 cmH2O/L, p=0.04), frequency dependence of resistance (1.02 ± 0.68 vs 0.05 ± 0.18 cmH2O/L/s, p=0.04), and resonance frequency (2.80 ± 4.14 vs -1.42 ± 2.13 cmH2O/L/s, p=0.04). Conclusion: Patients with asthma and co-existing OSA had greater overnight accumulation of fluid in the thorax in association with greater small airways narrowing than those without OSA. This suggests OSA could contribute to worsening of asthma at night by increasing fluid accumulation in the thorax.

Sleep apnea severity based on estimated tidal volume and snoring features from tracheal signals.

Sleep apnea can be characterized by reductions in the respiratory tidal volume. Previous studies showed that the tidal volume can be estimated from tracheal sounds and movements called tracheal signals. Additionally, tracheal sounds include the sounds of snoring, a common symptom of obstructive sleep apnea. This study investigates the feasibility of estimating the severity of sleep apnea, as quantified by the apnea/hypopnea index (AHI), using the estimated tidal volume and snoring sounds extracted from tracheal signals. Tracheal signals were recorded simultaneously with polysomnography (PSG). The tidal volume was estimated from tracheal signals. The reductions in the tidal volume were detected as potential respiratory events. Additionally, features related to snoring sounds, which quantified variability, temporal clusters, and dominant frequency of snores, were extracted. A step-wise regression model and a greedy search algorithm were used sequentially to select the optimal set of features to estimate the apnea/hypopnea index and classify participants into healthy individuals and patients with sleep apnea. Sixty-one participants with suspected sleep apnea (age: 51 ± 16, body mass index: 29.5 ± 6.4 kg/m2 , apnea/hypopnea index: 20.2 ± 21.2 event/h) who were referred for a sleep test were recruited. The estimated apnea/hypopnea index was strongly correlated with the polysomnography-based apnea/hypopnea index (R2  = 0.76, p < 0.001). The accuracy of detecting sleep apnea for the apnea/hypopnea index cutoff of 15 events/h was 78.69% and 83.61% with and without using snore-related features. These findings suggest that acoustic estimation of airflow and snore-related features can provide a convenient and reliable method for screening of sleep apnea.

Relative tidal volume and respiratory airflow estimation using tracheal sound and movement during sleep.

Airflow is the reference signal to assess sleep respiratory disorders, such as sleep apnea. Previous studies estimated airflow using tracheal sounds in short segments with specific airflow rates, while requiring calibration or a few breaths for tuning the relationship between sound energy and airflow. Airflow-sound relationship can change by posture, sleep stage and airflow rate or tidal volume. We investigated the possibility of estimating surrogates of tidal volume without calibration in the adult sleep apnea population using tracheal sounds and movements. Two surrogates of tidal volume: thoracoabdominal range of sum movement and airflow level were estimated. Linear regression was used to estimate thoracoabdominal range of sum movement from sound energy and the range of movements. The sound energy lower envelope was found to correlate with airflow level. The agreement between reference and estimated signals was assessed by repeated-measure correlation analysis. The estimated tidal volumes were used to estimate the airflow signal. Sixty-one participants (30 females, age: 51 ± 16 years, body mass index: 29.5 ± 6.4 kg m-2 , and apnoea-hypopnea index: 20.2 ± 21.2) were included. Reference and estimated thoracoabdominal range of sum movement of whole night data were significantly correlated with the reference signal extracted from polysomnography (r = 0.5 ± 0.06). Similarly, significant correlations (r = 0.3 ± 0.05) were found for airflow level. Significant differences in estimated surrogates of tidal volume were found between normal breathing and apnea/hypopnea. Surrogate of airflow can be extracted from tracheal sounds and movements, which can be used for assessing the severity of sleep apnea and even phenotyping sleep apnea patients based on the estimated airflow shape.

Detection of Apnea Bradycardia from ECG Signals of Preterm Infants Using Layered Hidden Markov Model.

Apnea-bradycardia (AB) is a common complication in prematurely born infants, which is associated with reduced survival and neurodevelopmental outcomes. Thus, early detection or predication of AB episodes is critical for initiating preventive interventions. To develop automatic real-time operating systems for early detection of AB, recent advances in signal processing can be employed. Hidden Markov Models (HMM) are probabilistic models with the ability of learning different dynamics of the real time-series such as clinical recordings. In this study, a hierarchy of HMMs named as layered HMM was presented to detect AB episodes from pre-processed single-channel Electrocardiography (ECG). For training the hierarchical structure, RR interval, and width of QRS complex were extracted from ECG as observations. The recordings of 32 premature infants with median 31.2 (29.7, 31.9) weeks of gestation were used for this study. The performance of the proposed layered HMM was evaluated in detecting AB. The best average accuracy of 97.14 ± 0.31% with detection delay of - 5.05 ± 0.41 s was achieved. The results show that layered structure can improve the performance of the detection system in early detecting of AB episodes. Such system can be incorporated for more robust long-term monitoring of preterm infants.

Automatic Respiratory Phase Identification Using Tracheal Sounds and Movements During Sleep.

One of the most important signals to assess respiratory function, especially in patients with sleep apnea, is airflow. A convenient method to estimate airflow is based on analyzing tracheal sounds and movements. However, this method requires accurate identification of respiratory phases. Our goal is to develop an automatic algorithm to analyze tracheal sounds and movements to identify respiratory phases during sleep. Data from adults with suspected sleep apnea who were referred for in-laboratory sleep studies were included. Simultaneously with polysomnography, tracheal sounds and movements were recorded with a small wearable device attached to the suprasternal notch. First, an adaptive detection algorithm was developed to localize the respiratory phases in tracheal sounds. Then, for each phase, a set of morphological features from sound energy and tracheal movement were extracted to classify the localized phases into inspirations or expirations. The average error and time delay of detecting respiratory phases were 7.62% and 181 ms during normal breathing, 8.95% and 194 ms during snoring, and 13.19% and 220 ms during respiratory events, respectively. The average classification accuracy was 83.7% for inspirations and 75.0% for expirations. Respiratory phases were accurately identified from tracheal sounds and movements during sleep.

Apnea bradycardia detection based on new coupled hidden semi Markov model.

In this paper, a method for apnea bradycardia detection in preterm infants is presented based on coupled hidden semi Markov model (CHSMM). CHSMM is a generalization of hidden Markov models (HMM) used for modeling mutual interactions among different observations of a stochastic process through using finite number of hidden states with corresponding resting time. We introduce a new set of equations for CHSMM to be integrated in a detection algorithm. The detection algorithm was evaluated on a simulated data to detect a specific dynamic and on a clinical dataset of electrocardiogram signals collected from preterm infants for early detection of apnea bradycardia episodes. For simulated data, the proposed algorithm was able to detect the desired dynamic with sensitivity of 96.67% and specificity of 98.98%. Furthermore, the method detected the apnea bradycardia episodes with 94.87% sensitivity and 96.52% specificity with mean time delay of 0.73 s. The results show that the algorithm based on CHSMM is a robust tool for monitoring of preterm infants in detecting apnea bradycardia episodes. Graphical Abstract Apnea Bradycardia detection using Coupled hidden semi Markov Model from electrocardiography. In this model, a sequence of hidden states is assigned to each observation based on the effects of previous states of all observations.

Sleep/Wakefulness Detection Using Tracheal Sounds and Movements.

PURPOSE: The current gold standard to detect sleep/wakefulness is based on electroencephalogram, which is inconvenient if included in portable sleep screening devices. Therefore, a challenge in the portable devices is sleeping time estimation. Without sleeping time, sleep parameters such as apnea/hypopnea index (AHI), an index for quantifying sleep apnea severity, can be underestimated. Recent studies have used tracheal sounds and movements for sleep screening and calculating AHI without considering sleeping time. In this study, we investigated the detection of sleep/wakefulness states and estimation of sleep parameters using tracheal sounds and movements. MATERIALS AND METHODS: Participants with suspected sleep apnea who were referred for sleep screening were included in this study. Simultaneously with polysomnography, tracheal sounds and movements were recorded with a small wearable device, called the Patch, attached over the trachea. Each 30-second epoch of tracheal data was scored as sleep or wakefulness using an automatic classification algorithm. The performance of the algorithm was compared to the sleep/wakefulness scored blindly based on the polysomnography. RESULTS: Eighty-eight subjects were included in this study. The accuracy of sleep/wakefulness detection was 82.3±8.66% with a sensitivity of 87.8±10.8 % (sleep), specificity of 71.4±18.5% (awake), F1 of 88.1±9.3% and Cohen's kappa of 0.54. The correlations between the estimated and polysomnography-based measures for total sleep time and sleep efficiency were 0.78 (p<0.001) and 0.70 (p<0.001), respectively. CONCLUSION: Sleep/wakefulness periods can be detected using tracheal sound and movements. The results of this study combined with our previous studies on screening sleep apnea with tracheal sounds provide strong evidence that respiratory sounds analysis can be used to develop robust, convenient and cost-effective portable devices for sleep apnea monitoring.

Distinguishing Obstructive Versus Central Apneas in Infrared Video of Sleep Using Deep Learning: Validation Study.

BACKGROUND: Sleep apnea is a respiratory disorder characterized by an intermittent reduction (hypopnea) or cessation (apnea) of breathing during sleep. Depending on the presence of a breathing effort, sleep apnea is divided into obstructive sleep apnea (OSA) and central sleep apnea (CSA) based on the different pathologies involved. If the majority of apneas in a person are obstructive, they will be diagnosed as OSA or otherwise as CSA. In addition, as it is challenging and highly controversial to divide hypopneas into central or obstructive, the decision about sleep apnea type (OSA vs CSA) is made based on apneas only. Choosing the appropriate treatment relies on distinguishing between obstructive apnea (OA) and central apnea (CA). OBJECTIVE: The objective of this study was to develop a noncontact method to distinguish between OAs and CAs. METHODS: Five different computer vision-based algorithms were used to process infrared (IR) video data to track and analyze body movements to differentiate different types of apnea (OA vs CA). In the first two methods, supervised classifiers were trained to process optical flow information. In the remaining three methods, a convolutional neural network (CNN) was designed to extract distinctive features from optical flow and to distinguish OA from CA. RESULTS: Overnight sleeping data of 42 participants (mean age 53, SD 15 years; mean BMI 30, SD 7 kg/m2; 27 men and 15 women; mean number of OA 16, SD 30; mean number of CA 3, SD 7; mean apnea-hypopnea index 27, SD 31 events/hour; mean sleep duration 5 hours, SD 1 hour) were collected for this study. The test and train data were recorded in two separate laboratory rooms. The best-performing model (3D-CNN) obtained 95% accuracy and an F1 score of 89% in differentiating OA vs CA. CONCLUSIONS: In this study, the first vision-based method was developed that differentiates apnea types (OA vs CA). The developed algorithm tracks and analyses chest and abdominal movements captured via an IR video camera. Unlike previously developed approaches, this method does not require any attachment to a user that could potentially alter the sleeping condition.

Portable diagnosis of sleep apnea with the validation of individual event detection.

STUDY OBJECTIVE: To develop an algorithm for improving apnea hypopnea index (AHI) estimation which includes event by event validation and event duration estimation. The algorithm uses breathing sounds, respiratory related movements and blood oxygen saturation (SaO2). METHODS: Adults with suspected sleep apnea underwent overnight polysomnography (PSG) at Toronto Rehabilitations Institute. Simultaneously with PSG, breathing sounds and respiratory related movements were recorded over the suprasternal notch using the Patch. The Patch had a microphone and an accelerometer to record respiratory sounds and movement, respectively. First, we calculated the amount of drops in SaO2 from pulse oximeter. Subsequently, energy of breaths and accelerometer were extracted. Features were normalized, weighted, summed and passed through a threshold to estimate PatchAHI. PatchAHI was compared to the AHI obtained from PSG (PSGAHI). Furthermore, performance of event detection was evaluated using F1-score. Moreover, event duration difference between estimated and PSG-based events was compared. RESULTS: Data from 69 subjects were investigated. PatchAHI had high correlation with PSGAHI (r2 = 0.88). Considering a diagnostic AHI cut-off of ≥15, sensitivity and specificity were 91.42 ± 11.92% and 89.29 ± 7.62%, respectively. F1-score for individual event detection increased from 0.22 ± 0.10 for AHI≤5 to 0.72 ± 0.09 for AHI >30. Moreover, event duration difference between estimated events and PSG-based events was 5.33 ± 8.17 sec. CONCLUSION: Our proposed algorithm had high accuracy in estimating individual respiratory events during sleep. The algorithm can increase reliability of acoustic methods for diagnosis of sleep apnea at home.

Switching Kalman filter based methods for apnea bradycardia detection from ECG signals.

Apnea bradycardia (AB) is an outcome of apnea occurrence in preterm infants and is an observable phenomenon in cardiovascular signals. Early detection of apnea in infants under monitoring is a critical challenge for the early intervention of nurses. In this paper, we introduce two switching Kalman filter (SKF) based methods for AB detection using electrocardiogram (ECG) signal.The first SKF model uses McSharry's ECG dynamical model integrated in two Kalman filter (KF) models trained for normal and AB intervals. Whereas the second SKF model is established by using only the RR sequence extracted from ECG and two AR models to be fitted in normal and AB intervals. In both SKF approaches, a discrete state variable called a switch is considered that chooses one of the models (corresponding to normal and AB) during the inference phase. According to the probability of each model indicated by this switch, the model with larger probability determines the observation label at each time instant.It is shown that the method based on ECG dynamical model can be effectively used for AB detection. The detection performance is evaluated by comparing statistical metrics and the amount of time taken to detect AB compared with the annotated onset. The results demonstrate the superiority of this method, with sensitivity and specificity 94.74[Formula: see text] and 94.17[Formula: see text], respectively. The presented approaches may therefore serve as an effective algorithm for monitoring neonates suffering from AB.