Noncontact Sleep Monitoring With Infrared Video Data to Estimate Sleep Apnea Severity and Distinguish Between Positional and Nonpositional Sleep Apnea: Model Development and Experimental Validation

Background Sleep apnea is a respiratory disorder characterized by frequent breathing cessation during sleep. Sleep apnea severity is determined by the apnea-hypopnea index (AHI), which is the hourly rate of respiratory events. In positional sleep apnea, the AHI is higher in the supine sleeping position than it is in other sleeping positions. Positional therapy is a behavioral strategy (eg, wearing an item to encourage sleeping toward the lateral position) to treat positional apnea. The gold standard of diagnosing sleep apnea and whether or not it is positional is polysomnography; however, this test is inconvenient, expensive, and has a long waiting list. Objective The objective of this study was to develop and evaluate a noncontact method to estimate sleep apnea severity and to distinguish positional versus nonpositional sleep apnea. Methods A noncontact deep-learning algorithm was developed to analyze infrared video of sleep for estimating AHI and to distinguish patients with positional vs nonpositional sleep apnea. Specifically, a 3D convolutional neural network (CNN) architecture was used to process movements extracted by optical flow to detect respiratory events. Positional sleep apnea patients were subsequently identified by combining the AHI information provided by the 3D-CNN model with the sleeping position (supine vs lateral) detected via a previously developed CNN model. Results The algorithm was validated on data of 41 participants, including 26 men and 15 women with a mean age of 53 (SD 13) years, BMI of 30 (SD 7), AHI of 27 (SD 31) events/hour, and sleep duration of 5 (SD 1) hours; 20 participants had positional sleep apnea, 15 participants had nonpositional sleep apnea, and the positional status could not be discriminated for the remaining 6 participants. AHI values estimated by the 3D-CNN model correlated strongly and significantly with the gold standard (Spearman correlation coefficient 0.79, P<.001). Individuals with positional sleep apnea (based on an AHI threshold of 15) were identified with 83% accuracy and an F1-score of 86%. Conclusions This study demonstrates the possibility of using a camera-based method for developing an accessible and easy-to-use device for screening sleep apnea at home, which can be provided in the form of a tablet or smartphone app.

Download Full-text

Noncontact Sleep Monitoring With Infrared Video Data to Estimate Sleep Apnea Severity and Distinguish Between Positional and Nonpositional Sleep Apnea: Model Development and Experimental Validation (Preprint)

10.2196/preprints.26524 ◽

2020 ◽

Author(s):

Sina Akbarian ◽

Nasim Montazeri Ghahjaverestan ◽

Azadeh Yadollahi ◽

Babak Taati

Keyword(s):

Sleep Apnea ◽

Gold Standard ◽

Learning Algorithm ◽

Video Data ◽

Apnea Hypopnea Index ◽

Respiratory Disorder ◽

Sleeping Position ◽

Respiratory Events ◽

Infrared Video ◽

3D Cnn

BACKGROUND Sleep apnea is a respiratory disorder characterized by frequent breathing cessation during sleep. Sleep apnea severity is determined by the apnea-hypopnea index (AHI), which is the hourly rate of respiratory events. In positional sleep apnea, the AHI is higher in the supine sleeping position than it is in other sleeping positions. Positional therapy is a behavioral strategy (eg, wearing an item to encourage sleeping toward the lateral position) to treat positional apnea. The gold standard of diagnosing sleep apnea and whether or not it is positional is polysomnography; however, this test is inconvenient, expensive, and has a long waiting list. OBJECTIVE The objective of this study was to develop and evaluate a noncontact method to estimate sleep apnea severity and to distinguish positional versus nonpositional sleep apnea. METHODS A noncontact deep-learning algorithm was developed to analyze infrared video of sleep for estimating AHI and to distinguish patients with positional vs nonpositional sleep apnea. Specifically, a 3D convolutional neural network (CNN) architecture was used to process movements extracted by optical flow to detect respiratory events. Positional sleep apnea patients were subsequently identified by combining the AHI information provided by the 3D-CNN model with the sleeping position (supine vs lateral) detected via a previously developed CNN model. RESULTS The algorithm was validated on data of 41 participants, including 26 men and 15 women with a mean age of 53 (SD 13) years, BMI of 30 (SD 7), AHI of 27 (SD 31) events/hour, and sleep duration of 5 (SD 1) hours; 20 participants had positional sleep apnea, 15 participants had nonpositional sleep apnea, and the positional status could not be discriminated for the remaining 6 participants. AHI values estimated by the 3D-CNN model correlated strongly and significantly with the gold standard (Spearman correlation coefficient 0.79, P<.001). Individuals with positional sleep apnea (based on an AHI threshold of 15) were identified with 83% accuracy and an F1-score of 86%. CONCLUSIONS This study demonstrates the possibility of using a camera-based method for developing an accessible and easy-to-use device for screening sleep apnea at home, which can be provided in the form of a tablet or smartphone app.

Download Full-text

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

10.2196/preprints.17252 ◽

2019 ◽

Author(s):

Sina Akbarian ◽

Nasim Montazeri Ghahjaverestan ◽

Azadeh Yadollahi ◽

Babak Taati

Keyword(s):

Sleep Apnea ◽

Optical Flow ◽

Central Sleep Apnea ◽

Video Camera ◽

Video Data ◽

Apnea Hypopnea Index ◽

Respiratory Disorder ◽

Obstructive Sleep ◽

Infrared Video ◽

Supervised Classifiers

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.

Download Full-text

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

Journal of Medical Internet Research ◽

10.2196/17252 ◽

2020 ◽

Vol 22 (5) ◽

pp. e17252

Author(s):

Sina Akbarian ◽

Nasim Montazeri Ghahjaverestan ◽

Azadeh Yadollahi ◽

Babak Taati

Keyword(s):

Sleep Apnea ◽

Optical Flow ◽

Central Sleep Apnea ◽

Video Camera ◽

Video Data ◽

Apnea Hypopnea Index ◽

Respiratory Disorder ◽

Obstructive Sleep ◽

Infrared Video ◽

Supervised Classifiers

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.

Download Full-text

Detecting Apnea/Hypopnea Events Time Location from Sound Recordings for Patients with Severe or Moderate Sleep Apnea Syndrome

Applied Sciences ◽

10.3390/app11156888 ◽

2021 ◽

Vol 11 (15) ◽

pp. 6888

Author(s):

Georgia Korompili ◽

Lampros Kokkalas ◽

Stelios A. Mitilineos ◽

Nicolas-Alexander Tatlas ◽

Stelios M. Potirakis

Keyword(s):

Sleep Apnea ◽

Sleep Apnea Syndrome ◽

Apnea Hypopnea Index ◽

Event Time ◽

Severity Estimation ◽

Apnea Duration ◽

Respiratory Events ◽

Moderate Sensitivity ◽

Apnea Syndrome

The most common index for diagnosing Sleep Apnea Syndrome (SAS) is the Apnea-Hypopnea Index (AHI), defined as the average count of apnea/hypopnea events per sleeping hour. Despite its broad use in automated systems for SAS severity estimation, researchers now focus on individual event time detection rather than the insufficient classification of the patient in SAS severity groups. Towards this direction, in this work, we aim at the detection of the exact time location of apnea/hypopnea events. We particularly examine the hypothesis of employing a standard Voice Activity Detection (VAD) algorithm to extract breathing segments during sleep and identify the respiratory events from severely altered breathing amplitude within the event. The algorithm, which is tested only in severe and moderate patients, is applied to recordings from a tracheal and an ambient microphone. It proves good sensitivity for apneas, reaching 81% and 70.4% for the two microphones, respectively, and moderate sensitivity to hypopneas—approx. 50% were identified. The algorithm also presents an adequate estimator of the Mean Apnea Duration index—defined as the average duration of the detected events—for patients with severe or moderate apnea, with mean error 1.7 s and 3.2 s for the two microphones, respectively.

Download Full-text

Predictors of changes in cerebral perfusion and oxygenation during obstructive sleep apnea

Scientific Reports ◽

10.1038/s41598-021-02829-4 ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Zhongxing Zhang ◽

Ming Qi ◽

Gordana Hügli ◽

Ramin Khatami

Keyword(s):

Obstructive Sleep Apnea ◽

Sleep Apnea ◽

Near Infrared ◽

Sleep Apnea Syndrome ◽

Apnea Hypopnea Index ◽

Mixed Effect ◽

Respiratory Events ◽

Obstructive Sleep ◽

Respiratory Event ◽

Titration Protocol

AbstractObstructive sleep apnea syndrome (OSAS) is a common sleep disorder. Severe OSAS defined as apnea–hypopnea index (AHI) ≥ 30/h is a risk factor for developing cerebro-cardiovascular diseases. The mechanisms of how repetitive sleep apneas/hypopneas damage cerebral hemodynamics are still not well understood. In this study, changes in blood volume (BV) and oxygen saturation (StO2) in the left forehead of 29 newly diagnosed severe OSAS patients were measured by frequency-domain near-infrared spectroscopy during an incremental continuous positive airway pressure (CPAP) titration protocol together with polysomnography. The coefficients of variation of BV (CV-BV) and the decreases of StO2 (de-StO2) of more than 2000 respiratory events were predicted using linear mixed-effect models, respectively. We found that longer events and apneas rather than hypopneas induce larger changes in CV-BV and stronger cerebral desaturation. Respiratory events occurring during higher baseline StO2 before their onsets, during rapid-eye-movement sleep and those associated with higher heart rate induce smaller changes in CV-BV and de-StO2. The stepwise increased CPAP pressures can attenuate these changes. These results suggest that in severe OSAS the length and the type of respiratory event rather than widely used AHI may be better parameters to indicate the severity of cerebral hemodynamic changes.

Download Full-text

Automatic detection of respiratory events during sleep from Polysomnography data using Layered Hidden Markov Model

Physiological Measurement ◽

10.1088/1361-6579/ac45e1 ◽

2021 ◽

Author(s):

Azadeh Sadoughi ◽

Mohammad Bagher Shamsollahi ◽

Emad Fatemizadeh

Keyword(s):

Sleep Apnea ◽

Probabilistic Models ◽

Markov Models ◽

Hidden Markov ◽

Automatic Detection ◽

Detection Algorithm ◽

Respiratory Disorder ◽

Respiratory Events ◽

Increased Risk ◽

Respiratory Event

Abstract Objective. Sleep apnea is a serious respiratory disorder, which is associated with increased risk factors for cardiovascular disease. Many studies in recent years have been focused on automatic detection of sleep apnea from polysomnography (PSG) recordings, however, detection of subtle respiratory events named Respiratory Event Related Arousals (RERAs) that do not meet the criteria for apnea or hypopnea is still challenging. The objective of this study was to develop automatic detection of sleep apnea based on Hidden Markov Models (HMMs) which are probabilistic models with the ability to learn different dynamics of the real time-series such as clinical recordings. Approach. In this study, a hierarchy of HMMs named Layered HMM was presented to detect respiratory events from PSG recordings. The recordings of 210 PSGs from Massachusetts General Hospital’s database were used for this study. To develop detection algorithms, extracted feature signals from airflow, movements over the chest and abdomen, and oxygen saturation in blood (SaO2) were chosen as observations. The respiratory disturbance index (RDI) was estimated as the number of apneas, hypopneas, and RERAs per hour of sleep. Main results. The best F1 score of the event by event detection algorithm was between 0.22±0.16 and 0.70±0.08 for different groups of sleep apnea severity. There was a strong correlation between the estimated and the PSG-derived RDI (R2=0.91, p<0.0001). The best recall of RERA detection was achieved 0.45±0.27. Significance. The results showed that the layered structure can improve the performance of the detection of respiratory events during sleep.

Download Full-text

Acute stroke and TIA patients have specific polygraphic features of obstructive sleep apnea

Sleep And Breathing ◽

10.1007/s11325-019-02010-2 ◽

2020 ◽

Vol 24 (4) ◽

pp. 1495-1505 ◽

Cited By ~ 2

Author(s):

Akseli Leino ◽

Susanna Westeren-Punnonen ◽

Juha Töyräs ◽

Sami Myllymaa ◽

Timo Leppänen ◽

...

Keyword(s):

Obstructive Sleep Apnea ◽

Ischemic Stroke ◽

Sleep Apnea ◽

Transient Ischemic Attack ◽

Sleep Disordered Breathing ◽

Apnea Hypopnea Index ◽

Stroke Patients ◽

Respiratory Events ◽

Obstructive Sleep ◽

Ischemic Attack

Abstract Purpose Obstructive sleep apnea (OSA) is associated with increased risk for stroke, which is known to further impair respiratory functions. However, it is unknown whether the type and severity of respiratory events are linked to stroke or transient ischemic attack (TIA). Thus, we investigate whether the characteristics of individual respiratory events differ between patients experiencing TIA or acute ischemic stroke and matched patients with clinically suspected sleep-disordered breathing. Methods Polygraphic data of 77 in-patients with acute ischemic stroke (n = 49) or TIA (n = 28) were compared to age, gender, and BMI-matched patients with suspected sleep-disordered breathing and no cerebrovascular disease. Along with conventional diagnostic parameters (e.g., apnea-hypopnea index), durations and severities of individual apneas, hypopneas and desaturations were compared between the groups separately for ischemic stroke and TIA patients. Results Stroke and TIA patients had significantly shorter apneas and hypopneas (p < 0.001) compared to matched reference patients. Furthermore, stroke patients had more central apnea events (p = 0.007) and a trend for higher apnea/hypopnea number ratios (p = 0.091). The prevalence of OSA (apnea-hypopnea index ≥ 5) was 90% in acute stroke patients and 79% in transient ischemic attack patients. Conclusion Stroke patients had different characteristics of respiratory events, i.e., their polygraphic phenotype of OSA differs compared to matched reference patients. The observed differences in polygraphic features might indicate that stroke and TIA patients suffer from OSA phenotype recently associated with increased cardiovascular mortality. Therefore, optimal diagnostics and treatment require routine OSA screening in patients with acute cerebrovascular disease, even without previous suspicion of OSA.

Download Full-text

Contactless recording of sleep apnea and periodic leg movements by nocturnal 3-D-video and subsequent visual perceptive computing

Scientific Reports ◽

10.1038/s41598-019-53050-3 ◽

2019 ◽

Vol 9 (1) ◽

Cited By ~ 2

Author(s):

Christian Veauthier ◽

Juliane Ryczewski ◽

Sebastian Mansow-Model ◽

Karen Otte ◽

Bastian Kayser ◽

...

Keyword(s):

Sleep Apnea ◽

Characteristic Curve ◽

Apnea Hypopnea Index ◽

Periodic Leg Movements ◽

Respiratory Events ◽

Obstructive Sleep ◽

Machine Learning Approach ◽

Diagnostic Benefit ◽

Test Probability ◽

Leg Movements

AbstractContactless measurements during the night by a 3-D-camera are less time-consuming in comparison to polysomnography because they do not require sophisticated wiring. However, it is not clear what might be the diagnostic benefit and accuracy of this technology. We investigated 59 persons simultaneously by polysomnography and 3-D-camera and visual perceptive computing (19 patients with restless legs syndrome (RLS), 21 patients with obstructive sleep apnea (OSA), and 19 healthy volunteers). There was a significant correlation between the apnea hypopnea index (AHI) measured by polysomnography and respiratory events measured with the 3-D-camera in OSA patients (r = 0.823; p < 0.001). The receiver operating characteristic curve yielded a sensitivity of 90% for OSA with a specificity of 71.4%. In RLS patients 72.8% of leg movements confirmed by polysomnography could be detected by 3-D-video and a significant moderate correlation was found between PLM measured by polysomnography and by the 3-D-camera (RLS: r = 0.654; p = 0.004). In total, 95.4% of the sleep epochs were correctly classified by the machine learning approach, but only 32.5% of awake epochs. Further studies should investigate, if this technique might be an alternative to home sleep testing in persons with an increased pre-test probability for OSA.

Download Full-text

Influence of the right- versus left-sided sleeping position on the apnea-hypopnea index in patients with sleep apnea

Sleep And Breathing ◽

10.1007/s11325-011-0547-4 ◽

2011 ◽

Vol 16 (3) ◽

pp. 617-620 ◽

Cited By ~ 8

Author(s):

Ozcan Ozeke ◽

Ozcan Erturk ◽

Mutlu Gungor ◽

Serap Bılen Hızel ◽

Dilek Aydın ◽

...

Keyword(s):

Sleep Apnea ◽

Apnea Hypopnea Index ◽

Sleeping Position ◽

The Right

Download Full-text

MONITORING OF NON-INVASIVE VITAL SIGNS FOR DETECTION OF SLEEP APNEA

Journal of Mechanics in Medicine and Biology ◽

10.1142/s0219519421400078 ◽

2021 ◽

pp. 2140007

Author(s):

HAN ZHANG ◽

WEIWEI ZHU ◽

SONGBIN YE ◽

SIHUA LI ◽

BAOXIAN YU ◽

...

Keyword(s):

Sleep Apnea ◽

Gold Standard ◽

Vital Signs ◽

Force Sensor ◽

Respiratory Disorder ◽

Morphological Variations ◽

Non Invasive ◽

Clinical Research Center ◽

Respiratory Signal

Sleep apnea (SA) syndrome is a respiratory disorder that occurs during the sleep. Polysomnography (PSG) has been widely applied by clinicians as a gold standard in the clinical diagnosis of SA syndrome. However, the use of PSG is inconvenient, intrusive, and significantly affects the sleep quality of patient. In this paper, we provide a nonintrusive solution for SA detection. Specifically, a force sensor was employed for the noninvasive vital sign acquisition during the patient’s sleep, where the respiratory signal was extracted adaptively by using the morphological filter. It was observed that the morphological variations before and during the occurrence of the SA events were significant for the SA discrimination. By taking advantage of the differential features with respect to the respiratory signal, the recognition of the SA event was performed using classifiers. For validation, the all-night PSG recordings of 12 volunteers with 8 SA syndrome patients were obtained from the National Clinical Research Center for Respiratory Disease. Numerical results showed that the proposed scheme achieved an averaged accuracy, sensitivity and specificity of 83.67%, 58.57% and 85.13%, respectively, for the SA recognition.

Download Full-text