Click publication for abstract and additional information. This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author’s copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.
Recent developments in personal and mobile healthcare have shown promising results in term of patients’ quality of life and quality of care improvements. This can be achieved through continuous monitoring of patients’ physiological functions using wearable non-invasive biomedical sensors. The remote collection and processing of such data can then be used to provide rapid medical response if a problem is detected or to offer preventive measures. However, the integration of wearable sensors into wider-scale framework is still a major challenge, as real-time data collection and remote configuration capabilities must be integrated to strongly constrained devices. Here, we show how such requirements can be integrated into a multiparameter, cardiorespiratory wearable sensor and how this sensor can be integrated into wide-scale Internet-based frameworks. We thus manufactured a biomedical-grade heart rate, instantaneous heart-rate variability and respiratory sensor. The sensor was tested in r eal life ambulatory condition, and we showed an Internet-based proof of concept exhibiting the integration of our sensor into wide-scale healthcare frameworks. Finally, we anticipate that wearable healthcare will greatly improve patients’ quality-of-life by using IoT-based wearable devices similar to the sensor developed in this paper.
The impressive evolution of neural networks and deep learning techniques during the last few years has offered new incomparable routes to solve many complex problems. Moreover, the fact that neural networks are structured and supervised has made it possible to perform automatic parameter tuning that guarantees convergence to the best expressive model for the problem assessed. In this work, we investigated the use of recurrent neural networks (RNNs) to solve the sequential sparse recovery problem through unfolding the iterative soft thresholding algorithm (ISTA) into a stacked RNN. Specifically, we examined the performance of the unsupervised iterative algorithm and the supervised network for a purely compressive sampling reconstruction problem of time-frequency representations. Our results demonstrated that the trained stacked neural network outperforms the iterative algorithm in the quality of the reconstructed data and points to several future directions to improve the performance.
In this paper we considered windows used for local vertex spectrum analysis of graph signals. In addition to a review of the convolution based windowing method, two methods based on the vertex neighborhood are presented. They are based on the graph path lengths. In the first one the number of edges in a path determine window size, while the edge weights are taken into account in the second method. Signal localization is performed by using these window functions. Windowing methods are used for signal local vertex spectrum calculation with a test signal. Norm one based concentration measure is used for comparison.
A method for a resistive circuit analysis based on graph spectral decompositions is proposed. It is shown that the Laplacian matrix can be used in order to calculate node potentials. Based on the Laplacian eigenvalues and eigenvectors it is possible to decompose a complex resistive circuit into smaller, weakly connected sub-circuits.
Electrodermal Activity (EDA) – a peripheral index of sympathetic nervous system activity - is a primary measure used in psychophysiology. EDA is widely accepted as an indicator of physiological arousal, and it has been shown to reveal when psychologically novel events occur. Traditionally, EDA data is collected in controlled laboratory experiments. However, recent developments in wireless biosensing have led to an increase in out-of-lab studies. This transition to ambulatory data collection has introduced challenges. In particular, artifacts such as wearer motion, changes in temperature, and electrical interference can be misidentified as true EDA responses. The inability to distinguish artifact from signal hinders analyses of ambulatory EDA data. Though manual procedures for identifying and removing EDA artifacts exist, they are time consuming – which is problematic for the types of longitudinal data sets represented in modern ambulatory studies. This manuscript presents a novel technique to automatically identify and remove artifacts in EDA data using curve fitting and sparse recovery methods. Our method was evaluated using labeled data to determine the accuracy of artifact identification. Procedures, results, conclusions, and future directions are presented.
In this paper, the use of mutual information and the Learn++.NSE algorithm is proposed to create an EEG SSVEP BCI system that can select and utilize data sets originating from a group of users. In typical BCI systems, the nonstationarity in the EEG prevents the system from blindly applying training data from other users to the incoming data. Mutual information is introduced to select previous data sets that provide the most information about current random variables. A signed rank test was employed to show that this configuration outperformed both normal Learn++.NSE ensembles and LDA classifiers. This indicates that mutual information and ensemble learning techniques may prove useful in improving user transferability in SSVEP systems with low computational requirements.
Brain-computer interfaces (BCIs) promise to promote a novel access channel for functional independence for individuals with severe speech and physical impairment (SSPI) that can occur as a result of numerous neurological diseases and injuries. Current BCI systems lack the robustness and accuracy to allow individuals with SSPI to complete tasks required for independent living (e.g. communication or navigation). We aim to develop a noninvasive hybrid BCI relying on two imaging modalities: Electroencephalography (EEG) and functional transcranial Doppler sonography (fTCD). Such hybrid BCI is expected to be sufficiently robust and accurate to be operated in a real-life environment.
The power law in the frequency spectrum S(f) = 1/fβ allows for a good representation of the various time evolution and complex interactions of many physiological processes. The spectral exponent β can be interpreted as the degree of fractal characteristic which in turn makes it some sort of biomarker that gives an idea of the relative health of an individual. The prediction of the 1/fβ time series can thus prove to be an asset in the medical field where forecasting the future health state of an individual can be important for rehabilitation purposes. The goal of this paper is to consider the accuracy of several time series prediction methods such as the neural networks, regression trees and bagged regression trees learning method. To test these methods we simulate stride intervals time series as 1/fβ processes. Our results show that the regression trees can accurately predict between five and fifteen points.
Wireless transcutaneous power transfer and communication has the potential to reduce the size of implantable medical devices, thereby reducing patient discomfort and minimizing the tissue area exposed to foreign material. Electromagnetic transmission mechanisms through tissue are determined by tissue structure and associated frequency-dependent tissue properties, which are significant in the design of wireless implantable medical devices. The purpose of this study was to investigate the effects of varying tissue dielectric properties on maximum power transfer to a subcutaneously implanted device in a paired electrode system designed for use in proximity to metallic orthopedic implants. The transcutaneous system including external and implanted electrode pairs was simulated at several radio frequencies (125 kHz, 1 MHz, 13.56 MHz, 403 MHz, and 915 MHz) while varying the dielectric properties of the tissue medium over a range of physiological values. Maximum power transfer was calculated to represent the best-case power gain across the range of tissue properties and frequencies, and greater achievable efficiencies were seen with higher quality factor as a function of the tissue properties. The results suggest that in the paired electrode system, utilization of capacitive coupling allows the system to function in proximity to metallic surfaces such as orthopedic implants. The results also suggest that higher power gains are possible through a choice of implant location based on expected tissue properties.
Acquiring swallowing accelerometry signals using a comprehensive sensing scheme may be a desirable approach for monitoring swallowing safety for longer periods of time. However, it needs to be insured that signal characteristics can be recovered accurately from compressed samples. In this paper, we considered this issue by examining the effects of the number of acquired compressed samples on the calculated swallowing accelerometry signal features. We used tri-axial swallowing accelerometry signals acquired from seventeen stroke patients (106 swallows in total). From acquired signals, we extracted typically considered signal features from time, frequency and time-frequency domains. Next, we compared these features from the original signals (sampled using traditional sampling schemes) and compressively sampled signals. Our results have shown we can obtain accurate estimates of signal features even by using only a third of original samples.
Emerging methods for the spectral analysis of graphs are analyzed in this paper, as graphs are currently used to study interactions in many fields from neuroscience to social networks. There are two main approaches related to the spectral transformation of graphs. The first approach is based on the Laplacian matrix. The graph Fourier transform is defined as an expansion of a graph signal in terms of eigenfunctions of the graph Laplacian. The calculated eigenvalues carry the notion of frequency of graph signals. The second approach is based on the graph weighted adjacency matrix, as it expands the graph signal into a basis of eigenvectors of the adjacency matrix instead of the graph Laplacian. Here, the notion of frequency is then obtained from the eigenvalues of the adjacency matrix or its Jordan decomposition. In this paper, advantages and drawbacks of both approaches are examined. Potential challenges and improvements to graph spectral processing methods are considered as well as the generalization of graph processing techniques in the spectral domain. Its generalization to the time-frequency domain and other potential extensions of classical signal processing concepts to graph datasets are also considered. Lastly, it is given an overview of the compressive sensing on graphs concepts.
PDF | DOI:
Today we are faced with what some call the “automation paradox” and others call “the ironies of automation”. Lisanne Bainbridge (1983) cautioned that the more automated a system becomes, the more important it is to appropriately integrate human contributions into the system. There is no question that automated control systems provide immeasurable benefit (improved efficiency, reliability, accuracy, safety, etc.); however, this comes at a cost; loss of skill, knowledge, decision-making capability and reaction-time in our human operators. Without daily engagement in the cognitive performance-based activities required by a control system, humans become less useful to the system. There seem to be two prevailing schools of thought on the best approach to human-system interface design. One advocates automating the system as much as possible to keep the human operator out of harm’s way and to remove the error-prone human from critical operations. The other camp claims that the human operator suffers significant losses in physical capability, memory and attention capacity and their learned responses diminish in quality if they have not been actively and cognitively engaged in the operation. Then when called upon to take over control in an automated system, they are less capable of effectively operating the system manually. It is not well enough established which is correct or if there is a generalizable correct path. Currently, as automated control systems are designed, it is often the case that operators are left in the operational fringes. So, researchers at the University of Pittsburgh have been exploring design methodologies that will allow system designers to understand and to then implement the best of the control system and human performance attributes, with the intent of concurrently minimizing the likelihood of human error-induced incidents. Through a series of trials using licensed reactor operators in a reconfigurable control room simulator, researchers are planning to identify and measure key performance variables while varying task configuration, level of automation, and override authority. If impact on system output is shown to be predictable, the generated model can help designers simulate various design strategies and their resulting impact on system performance, thus providing more-informed training protocols and content, more-informed simulator practice decisions and improved operational and operating procedure consistency.
Swallowing accelerometry is a promising tool for non-invasive assessment of swallowing difficulties. A recent contribution showed that swallowing accelerometry signals for healthy swallows and swallows indicating laryngeal penetration or tracheal aspiration have different time-frequency structures, which may be problematic for compressive sensing schemes based on time-frequency dictionaries. In this paper, we examined the effects of different swallows on the accuracy of a compressive sensing scheme based on modulated discrete prolate spheroidal sequences. We utilized tri-axial swallowing accelerometry signals recorded from four patients during routinely schedule videofluoroscopy exams. In particular, we considered 77 swallows approximately equally distributed between healthy swallows and swallows presenting with some penetration/aspiration. Our results indicated that the swallow type does not affect the accuracy of a considered compressive sensing scheme. Also, the results confirmed previous findings that each individual axis contributes different information. Our findings are important for further developments of a device which is to be used for long-term monitoring of swallowing difficulties.
Analog sparse signals resulting from biomedical and sensing network applications are typically non–stationary with frequency–varying spectra. By ignoring that the maximum frequency of their spectra is changing, uniform sampling of sparse signals collects unnecessary samples in quiescent segments of the signal. A more appropriate sampling approach would be signal–dependent. Moreover, in many of these applications power consumption and analog processing are issues of great importance that need to be considered. In this paper we present a signal dependent non–uniform sampler that uses a Modified Asynchronous Sigma Delta Modulator which consumes low–power and can be processed using analog procedures. Using Prolate Spheroidal Wave Functions (PSWF) interpolation of the original signal is performed, thus giving an asynchronous analog to digital and digital to analog conversion. Stable solutions are obtained by using modulated PSWFs functions. The advantage of the adapted asynchronous sampler is that range of frequencies of the sparse signal is taken into account avoiding aliasing. Moreover, it requires saving only the zero–crossing times of the non–uniform samples, or their differences, and the reconstruction can be done using their quantized values and a PSWF–based interpolation. The range of frequencies analyzed can be changed and the sampler can be implemented as a bank of filters for unknown range of frequencies. The performance of the proposed algorithm is illustrated with an electroencephalogram (EEG) signal.
Transcranial Doppler sonography was recently proposed as an approach for brain-machine interfaces. However, monitoring maximal cerebral blood flow velocity signals for extensive time periods can generate large volumes of data for processing. In this paper, a compressive sensing (CS) approach is proposed based on a time-frequency dictionary formed by modulated discrete prolate spheroidal sequences (MDPSS). To test the proposed scheme, we examined maximal cerebral blood flow velocity signals acquired from 20 healthy subjects during a resting state. The results of our analysis clearly depicted that these signals can be accurately reconstructed using only 30% and 50% of original samples. Hence, the proposed MDPSS-based CS approach is a valid tool for diminishing the number of acquired samples during brain-machine operations using transcranial Doppler sonography.
A recent rise of compressive sensing (CS) algorithms has prompted many questions about the analysis of such sensed signals. Specifically, calculating a time-frequency representation (TFR) of these signals is an open question. In this paper, we propose an approach for calculating TFRs of compressed sensed signals based on recently proposed CS algorithm using modulated discrete prolate spheroidal sequences (MDPSS). The results of our numerical analysis show that a visually reliable TFR of compressed sensed signals can be obtained using the proposed approach. Furthermore, these compressed sensed signals can also be used for accurate estimation of signal parameters such as the instantaneous frequency.
This paper aims to develop a UHF Gen 2 RFID system for transcutaneous operation for identifying and monitoring orthopedic implants. The major problems of using existing RFID antennas at UHF band for transcutaneous operation include possible interference with pacemakers and interference with other RFID systems working at the same band. To provide a solution for the above problems, this paper uses transcutaneous near field communication (TNFC) technology based on capacitive coupling between the reader and the tag. The reader and tag both have two electrodes and the energy and signal transmission occurs with the coupling between the electrodes. Both the reader and the tag are impedance matched for maximum power transmission efficiency. A reading range of 4.1 cm is achieved through pork skin using 30 dBm power from the reader. The proposed UHF RFID system is a feasible solution to provide high efficiency for transcutaneous operation while can eliminate an interference with other medical devices or RFID devices.
The Internet of Things describes multiple distributed systems where all (or most) everyday items include embedded systems in order to connect to the internet. This paradigm has the potential to revolutionize global industry and daily life. Healthcare is once such industry where the Internet of Things may provide great advantages to patients, care givers, and medical institutions. As the number of radio frequency emitters increases under this new paradigm public health and safety must also be taken into account. This paper explores the electromagnetic interference on implantable cardiac rhythm management devices caused by RFID interrogators. A standard electromagnetic compatibility test framework is proposed in order to diagnose the possibility of interference. Also, a mitigation method is proposed and tested. It is shown that the proposed method can reduce the incidence of clinically significant interference by nearly 60%.
Asynchronous signal processing is an appropriate low-power approach for the processing of bursty signals typical in biomedical applications and sensing networks. Different from the synchronous processing, based on the Shannon- Nyquist sampling theory, asynchronous processing is free of aliasing constrains and quantization error, while allowing continuous-time processing. In this paper we connect level crossing sampling with time-encoding using asynchronous sigma delta modulators, to develop an asynchronous decomposition procedure similar to the Haar transform wavelet decomposition. Our procedure provides a way to reconstruct bounded signals, not necessarily band-limited, from related zero-crossings, and it is especially applicable to decompose sparse signals in time and to denoise them. Actual and synthetic signals are used to illustrate the advantages of the decomposer.
Because of the need to control power consumption, in many biomedical applications asynchronous processing of the data is more appropriate. In this paper, we present a scale-based decomposition algorithm for analog signals similar to the wavelet decomposition. Our procedure uses asynchronous sigma delta modulators (ASDMs) to represent the amplitude of a signal using the zero-crossing times of a binary signal. Changing the zero-crossing times into random sequences of pulse widths, it can be shown to be equivalent to an optimal level-crossing sampler using local averages as the quantization levels. Applying the generation of multi-level signals from the output of ASDMs for different scale parameters we are able to obtain a decomposer that in a few stages provides a close representation of the signal. To illustrate the performance of the proposed decomposition, we consider its application to the representation of heart sounds.
A unified approach for the estimation of the first three phase derivatives of non-stationary signals is proposed in this paper. The possibility to accurately estimate phase derivatives is important in many applications dealing with objects velocity, acceleration and acceleration rate, such as the radar applications and mechanics. The estimation approach is based on definition of the complex-lag distribution. The proposed distribution is inspired by the concepts of complex analysis theory. The general form of distribution for the estimation of the first, second and third derivative of the phase is derived from the specific individual cases. The theoretical considerations are illustrated in the example with fast varying signal phase function.
Continuous monitoring of physiological functions such as heart sounds can pose severe constraints on data acquisition and processing systems, especially if remote monitoring is desired. In this paper, we investigate the utility of a recently proposed compressive sensing (CS) algorithm based on modulated discrete prolate spheroidal sequences (MDPSS) for recovering sparsely sampled heart sounds. In particular, we investigate the recordings containing opening snap (OS) or the third heart sounds (S3) in addition to first and second heart sounds. The results of numerical analysis show that heart sounds can be accurately reconstructed even when the sampling rate is reduced to 40% of the original sampling frequency.
Continuous-time digital signal processors not only offer significant energy savings in important applications such as implantable biomedical devices, but can implement asynchronous procedures. In this paper, we propose an asynchronous signal decomposition for continuous-time signals based on scale rather than frequency. Because the implementation of the proposed procedure does not use a clock it is not affected by aliasing, and moreover no quantization is involved. Such procedure is specially applicable to biomedical signals delivering information in bursts rather than continuously. The decomposer consists of cascaded modules that expand the signal onto different resolution scales and each is composed of an asynchronous sigma delta modulator (ASDM) followed by a local averager and a lowpass filter. The ASDM is a non-linear feedback system used to represent the amplitude of a continuous-time signal by a binary signal whose zero-crossings are used to reconstruct the original signal. One of the parameters of the ASDM is used as a scaling parameter, permitting us to represent the signal by its local means –at different scales– and computed from the zerocrossing times of the output of the ASDM. We develop a compact signal representation that is described by a small number of scale parameters and contains information useful in the continuoustime processing and transmission of the data. The performance of the proposed procedure is illustrated using different types of signals. As a practical application, we consider the non-linear denoising of swallowing signals. Potentially our procedure will find application in asynchronous signal acquisition, continuoustime digital signal processing and transmission in low-power biomedical applications.
Breath sounds in patients with obstructive sleep apnea are very dynamic and variable signals due to their versatile nature. In this paper, we present an adaptive segmentation algorithm for these sounds. The algorithm divides the breath sounds into segments with similar amplitude levels. As the first step, the proposed scheme creates an envelope of the signal characterizing its long term amplitude variations. Then, K-means clustering is iteratively applied to detect borders between different segments in the envelope, which will then be used to segment and normalize the original signal.
The reassignment method is a widespread approach for obtaining high resolution time-frequency representations. Nevertheless, its performance is not always optimal and can deteriorate for low signal-to-noise ratio (SNR) values. In order to overcome these obstacles, a novel method for obtaining high resolution time-frequency representations is proposed in this paper. The new method implements proposed nonparametric snakes in order to obtain accurate locations of the signal ridges in the time-frequency domain. The results of numerical analysis show that the proposed method is capable of achieving significantly higher concentration of signals in the time-frequency domain in comparison to the spectrogram and the traditional reassignment method. Furthermore, the new scheme also maintains good performance for low SNR values, while the performance of the other two considered methods significantly diminishes. It is clear from the results that the proposed method might be of significance in applications where accurate estimation of the signal components is required for low SNR values.
The paper presents two novel applications of Thomson Multitaper Analysis. It is shown how a wideband simulator of a double mobile MIMO channel could be developed based on geometrical channel model. It is also shown how modification of Discrete Prolate Spheroidal Sequences could be used to better estimation of sparse channels. A number of other potential applications is also mentioned.
We present a non-invasive brain-machine-interface (BMI) prototype system which allows the simple control of a switch. The main goal of the system, based on electroencephalogram (EEG) recordings, is to create mechanical action from brain activity. Experimental work presented in this paper outlines the operation of a system which is a crude imitation of an ultrasound echolocation based vision mechanism, commonly used by bats and dolphins, which is controlled by brain activity. Simple time-frequency-space domain signal analysis methods are employed to generate the electrical control-signal, while the sonar transducer is mounted on a robotic arm capable of scanning the upper hemisphere.
Accurate and sparse representation of a moderately fast fading channel using bases functions is achievable when both channel and bases bands align. If a mismatch exists, usually a larger number of bases functions is needed to achieve the same accuracy. In this paper, we propose a novel approach for channel estimation based on frames, which preserves sparsity and improves estimation accuracy. Members of the frame are formed by modulating and varying the bandwidth of discrete prolate spheroidal sequences (DPSS) in order to reflect various scattering scenarios. To achieve the sparsity of the proposed representation, a matching pursuit approach is employed. The estimation accuracy of the scheme is evaluated and compared with the accuracy of a Slepian basis expansion estimator based on DPSS for a variety of mobile channel parameters. The results clearly indicate that for the same number of atoms, a significantly higher estimation accuracy is achievable with the proposed scheme when compared to the DPSS estimator.
Clinical experience has shown that heart sounds can be an effective tool to noninvasively diagnose some forms of heart disease. In this paper, an algorithm based on time-frequency analysis is used for the decomposition of the heart sounds. The decomposition algorithm is based on the S-method. The S-method is a time-frequency representation that can produce a distribution equal or close to the sum of the Wigner distributions of individual signal components. The decomposition algorithm is used for segmentation of the heart sound recordings that contain either of the two sounds: opening snap or third heart sound, which indicate distinct heart diseases. The results show that the algorithm effectively decomposes each heart beat into the corresponding components. Hence, they may be used in conjunction with a classification algorithm, allowing automatic decomposition and classification of the heart diseases associated with the opening snap and the third heart sound.
A time-frequency signal analysis tool, known as S-transform, can suffer from poor energy concentration in the time-frequency domain. In this paper, a frequency dependent Kaiser window is presented for improving the energy concentration of the S-transform. The new window is analyzed using a set of test signals. The results indicate that the proposed scheme can significantly improve the energy concentration in the time-frequency domain in comparison with the standard S-transform.
In this paper, it is demonstrated that time-frequency techniques may be used for an enhanced diagnosis of the heart disease. Three time-frequency analysis techniques are compared: spectrogram, Wigner distribution and Smethod. The results show that the S-method could be used in the heart sounds analysis, and that is also capable of enhancing the diagnostic techniques available to medical personnel.
The effect of three time-frequency representations on a novel correlation algorithm is studied. By representing a signal in the time-frequency domain, a redundant representation of the signal is obtained. The algorithm presented relies on such redundancies to extrapolate some significant features of the signal. The developed scheme has been applied to heart sound analysis using real recordings from patients, where the opening snap (OS) is distinguished from the third heart sound (S3). The results for the three time-frequency transforms are compared and very encouraging results have been obtained with S-transform.