Takahiro Emoto

Artificial neural networks for breathing and snoring episode detection in sleep sounds.

Obstructive sleep apnea (OSA) is a serious disorder characterized by intermittent events of upper airway collapse during sleep. Snoring is the most common nocturnal symptom of OSA. Almost all OSA patients snore, but not all snorers have the disease. Recently, researchers have attempted to develop automated snore analysis technology for the purpose of OSA diagnosis. These technologies commonly require, as the first step, the automated identification of snore/breathing episodes (SBE) in sleep sound recordings. Snore intensity may occupy a wide dynamic range (> 95 dB) spanning from the barely audible to loud sounds. Low-intensity SBE sounds are sometimes seen buried within the background noise floor, even in high-fidelity sound recordings made within a sleep laboratory. The complexity of SBE sounds makes it a challenging task to develop automated snore segmentation algorithms, especially in the presence of background noise. In this paper, we propose a fundamentally novel approach based on artificial neural network (ANN) technology to detect SBEs. Working on clinical data, we show that the proposed method can detect SBE at a sensitivity and specificity exceeding 0.892 and 0.874 respectively, even when the signal is completely buried in background noise (SNR < 0 dB). We compare the performance of the proposed technology with those of the existing methods (short-term energy, zero-crossing rates) and illustrate that the proposed method vastly outperforms conventional techniques.

High frequency region of the snore spectra carry important information on the disease of sleep apnoea

Snoring is the most common symptom of obstructive sleep apnoea (OSA). Several researchers have reported differences between the power spectra of non-OSA and OSA snorers. The traditional approach over the years has been to record snore sounds at a bandwidth of < 5 kHz. Narrowing of the upper airways during OSA events and the resulting upward shift of snore frequencies also lend support to the idea of examining snore sounds beyond 5 kHz. In this paper, we compute the power spectra of snores in three different bands defined as: low-frequency band (LFB: < 5 kHz); middle-frequency band (MFB: 5–10 kHz) and high-frequency band (HFB: 10–20 kHz). We illustrate that there is a significant difference between non-OSA snorers (Apnoea Hypopnoea Index (AHI) < 10) and OSA snorers (AHI > 10) in the region > 5 kHz. We then develop a feature to diagnose OSA based on the spectral differences in the high frequency region and evaluate its performance on a database of 20 subjects. Our results strongly suggest that the high-frequency region of the snore sounds carry information, hitherto disregarded, on the disease of sleep apnoea.

Feature Extraction for Snore Sound via Neural Network Processing

Snore sound (SS) is the earliest and the most common symptom of Obstructive Sleep Apnea (OSA) which is a serious disease caused by the collapse of upper airways during sleep. SS should carry vital information on the state of the upper airways and is simple to acquire and rich in features but their analysis is complicated. In this study we use neural network (NN) based method to model SS via a simple second order one-step predictor. We show that the some hidden information/feature of a SS can be conveniently captured in the connection-weight-space (CWS) of the NN, after a process of supervised training. The availability of the proposed method is investigated by performing independent component analysis (ICA) on CWS.

Evaluation of blood flow velocity waveform in common carotid artery using multi-branched arterial segment model of human arteries

Abstract Arteriosclerosis is considered to be a major cause of cardiovascular diseases, which account for approximately 30% of the causes of death in the world. We have recently demonstrated a strong correlation between arteriosclerosis (arterial elasticity) and two characteristics: maximum systolic velocity (S1) and systolic second peak velocity (S2) of the common carotid artery flow velocity waveform (CCFVW). The CCFVW can be measured by using a small portable measuring device. However, there is currently no theoretical evidence supporting the causes of the relation between CCFVW and arterial elasticity, or the origin of the CCFVW characteristics. In this study, the arterial blood flow was simulated using a one-dimensional systemic arterial segments model of human artery in order to conduct a qualitative evaluation of the relationship between arterial elasticity and the characteristics of CCFVW. The simulation was carried out based on the discretized segments with the physical properties of a viscoelastic tube (the cross-sectional area at the proximal and terminal ends, the length, and the compliance per unit area of the tube (CS)). The findings obtained through this study revealed that the simulated CCFVW had shape similar characteristics to that of the measured CCFVW. Moreover, when the compliance CS of the model was decreased, the first peak of the simulated-CCFVW decreased and the second peak increased. Further, by separating the anterograde pulse wave and the reflected pulse wave, which form the CCFVW, we found that the decrease in the first peak of the simulated CCFVW was due to the arrival of a reflected pulse wave from the head after the common carotid artery toward the arrival of a anterograde pulse wave ejected directly from the heart and that the increase in the second peak resulted from the arrival of the peak of the reflected pulse wave from the thoracic aorta. These results establish that the CCFVW characteristics contribute to the assessment of arterial elasticity.

Analysis of a nonlinear system via internal-state of a neural network

Summary form only given. The internal state of a network has been inspected to evaluate the performance of the network. In particular, the weight vectors of the network have been applied for the analysis of a time series such as biological signals and nonstationary signals. The complexity (eg. nonlinearity and nonstationarity) of such signals often makes it a challenging task to use them in the signal processing field. In this paper, we propose a new neural network based technique to address these problems. We show that a feed forward, multi-layered neural network can conveniently capture the parameter change of a nonlinear system in its connection weight-space, after a process of supervised training. The performance of the proposed method is investigated with a linear and nonlinear system simulated via a mathematical equation.

Evaluation of human bowel motility using non-contact microphones

In recent years, features extracted from bowel sound (BS) has been proposed in the evaluation and monitoring of gastrointestinal motility. BS has been acquired using sensors such as electronic stethoscopes which require body contact with a subject. However, to our knowledge, the analysis of BS using non-contact microphones has not been reported yet. In this study, we compared BS acquired from a non-contact microphone with those of a stethoscope, while subjects undergo the soda tolerance test (STT). The STT stimulates bowel motility. Our investigation shows that irrespective of the body mass index, both stethoscope-based and non-contact microphone-based BSs have a similar performance in computing features such as: BS detected per minute and sound to sound interval in the time domain. These findings provide a powerful new window into the non-contact evaluation of bowel motility.

Neural networks for snore sound modeling in sleep apnea

Snoring is the earliest and the most common symptom of Obstructive Sleep Apnea (OSA) which is a serious disease caused by the collapse of upper airways during sleep. Recently, a few pioneering attempts have been made to use snore sounds (SS) is diagnosing OSA. The SS are simple to acquire and rich in features but their analysis is complicated. In this paper, we propose a neural network (NN) based method to model SS via a technique associated with k-step prediction. We also show that the features of a SS can be conveniently captured in the connection-weight-space (CWS) of the NN, after a process of supervised training. The performance of the proposed method is investigated via simulated and clinically measured data.

Blue light-emitting diodes induce autophagy in colon cancer cells by Opsin 3

Light emitting‐diodes (LED) have various effects on living organisms and recent studies have shown the efficacy of visible light irradiation from LED for anticancer therapies. However, the mechanism of LEDs effects on cancer cells remains unclear. The aim of the present study was to investigate the effects of LED on colon cancer cell lines and the role of photoreceptor Opsin 3 (Opn3) on LED irradiation in vitro.

Automatic snore sound extraction from sleep sound recordings via auditory image modeling

One of humans’ auditory abilities is differentiation between sounds with slightly different frequencies. Recently, the auditory image model (AIM) was developed to numerically explain this auditory phenomenon. Acoustic analyses of snore sounds have been performed recently by using non-contact microphones. Snore/non-snore classification techniques have been required at the front-end of snore analyses. The performances of sound classification methods can be evaluated based on human hearing, which is considered to be the gold standard. In this paper, we propose a novel method of automatically extracting snore sounds from sleep sounds by using an AIM-based snore/non-snore classification system. We report that the proposed automatic classification method could achieve a sensitivity of 97.2% and specificity of 96.3% when analyzing snore and non-snore sounds from 40 subjects. It is anticipated that our findings will contribute to the development of an automated snore analysis system to be used in sleep studies.

Theoretical study of evaluation method for MRI metal artifact

In dental field, the effect of the Magnetic Resonance Imaging (MRI) artifact generated by the magnetic metal is a significant problem. The MRI metal artifact occurs when using magnetic attachment and the keeper of the ferromagnetic substance remains implanted in the mouth as the MR image is taken. In this study, we theoretically evaluated and analyzed the artifact of MR images caused by the keeper based on the actual principle of MRI by means of simulation. As a result we were able to recognize the changes and distortion in the signal strength of the output image. We found that our results of output images and previously reported results of actual measurement are very similar. MRI artifact caused by dental magnetic metal showed that it can be reported by theoretical simulation.