E. Hari Krishna

Evaluation of wavelets for reduction of motion artifacts in photoplethysmographic signals

Photoplethysmographic (PPG) signals obtained at Red and Infrared wavelengths are utilized in pulse oximetry for estimation of arterial blood oxygen saturation (SpO2). Mostly inaccurate readings in a pulse oximeter arise when PPG signals are contaminated with motion artifacts (MA) due to the movement of patient and hence MA are a common cause of oximeter failure and loss of accuracy. This paper presents performance evaluation of different wavelets for the reduction of MA. Test results on the PPG signals recorded with frequently encountered artifacts (viz., horizontal, vertical and bending motions of finger) reveal that the estimated Spo2 values from MA reduced PPGs using different wavelets are very close to each other. In addition, the Daubechies wavelet interestingly kept the respiratory information intact while effectively removing the MA from PPG signals. Hence, the work establishes that the Daubechies wavelet is the most preferred wavelet for pulse oximetry applications.

Estimation of respiration rate from ECG, BP and PPG signals using empirical mode decomposition

Estimation of respiration rates from electrocardiogram (ECG), blood pressure (BP) and photoplethysmographic (PPG) signals would be an alternative approach for obtaining respiration related information. This process is useful in situations when, ECG, BP and PPG but not respiration is routinely monitored or in cases where, the cardiac arrhythmias are to be studied in correlation with respiratory information and is extremely important. There have been several efforts on ECG-Derived Respiration (EDR), BP-Derived Respiration (BDR) and PPG Derived Respiration (PDR). These methods are based on different signal processing techniques like filtering, wavelets and other statistical methods, which work by extraction of respiratory trend embedded into various physiological signals, as an additive component, or an amplitude modulated (AM) component or frequency modulated (FM) component. The proposed method is a robust, yet simple and makes use of derived Intrinsic Mode Functions (IMF) using Empirical Mode Decomposition (EMD). Test results on ECG, BP and PPG signals of the well known MIMIC database from Physiobank archive reveal that the proposed EMD method has efficiently extracted respiratory information from ECG, BP and PPG signals. The evaluated similarity parameters in both time and frequency domains for original and estimated respiratory rates have shown the superiority of the method.

Robust Extraction of Respiratory Activity From PPG Signals Using Modified MSPCA

The pulse oximeters photoplethysmographic (PPG) signals can be well utilized for extracting the vital respiratory activity, in addition to saturation and heart rate estimations, avoiding the usage of additional sensor for recording respiratory signal, in turn reducing the number of sensors connected to the patients body for recording vital signals. In this paper, we present a robust algorithm called modified multi scale principal component analysis (MMSPCA), for extraction of respiratory activity embedded in the PPG signals. The PPG signals are more commonly corrupted by motion artifacts (MA) due to voluntary or involuntary movements of the patients, making it difficult for the algorithms to extract respiratory signals. The problem of extracting respiratory signals from PPGs in the presence of MA is addressed for the first time in this paper. The problem gets aggravated when PPGs are severely afflicted with MAs in situations such as the MA frequency band (usually below 0.2 Hz) overlapping on to the band of respiratory frequencies (0.2-0.4 Hz). In the presented algorithm, the kurtosis and energy contribution levels (ECL) of approximate and detail coefficients are calculated for each wavelet sub-band matrix, generating a modified wavelet sub-band matrix. This makes the presented algorithm based on MMSPCA more robust in the sense that it is made motion resistant by suitably modifying the MSPCA. Functioning of the proposed algorithm is tested on the data recorded from 15 healthy subjects. Each data set consists of intentionally created possible MA noises, viz., vertical, horizontal, waving, and pressing MAs with different breathing patterns. The method is also applied on the recordings available with MIMIC database of Physionet archive. The statistical and error analysis, performed to test the efficacy of the presented MMSPCA algorithm, revealed a very good acceptance for derived respiratory signal, when compared with the originally recorded respiratory signals using classical method. The MMSPCA method clearly outperformed the conventional MSPCA-based method in the presence of MA.

On the Performance of Wavelets in Reducing Motion Artifacts from Photoplethysmographic Signals

In a pulse oximeter, clean artifact-free photoplethysmographic (PPG) signals with clearly separable DC and AC parts are necessary for error-free estimation of arterial oxygen saturation (SpO2). Motion artifacts (MA) introduced in the PPG signals due to the movement of a patient result in a significant error in the readings of pulse oximeters and hence are a common cause of oximeter failure and loss of accuracy. In this paper, we present a comparative analysis of using different wavelets for the case of MA reduction from corrupted PPG signals. PPG signals with frequently encountered artifacts (horizontal, vertical and bending motions of finger) were recorded from the subjects and processed with different wavelets. Simulation results and statistical analysis revealed that the Daubechies wavelet performed well for obtaining clean motion artifact-free PPG signal. Interestingly, the wavelet based method proved to be efficient in reducing MA restoring the respiratory information in tact with the PPG.

Acoustic echo cancellation using a computationally efficient transform domain LMS adaptive filter

Applications such as hands-free telephony, tele-classing and video-conferencing require the use of an acoustic echo canceller (AEC) to eliminate acoustic feedback from the loudspeaker to the microphone. Room acoustic echo cancellation typically requires adaptive filters with thousands of coefficients. Transform domain adaptive filter finds best solution for echo cancellation as it results in a significant reduction in the computational burden. Literature finds different orthogonal transform based adaptive filters for echo cancellation. In this paper, we present Hirschman Optimal Transform (HOT) based adaptive filter for elimination of echo from audio signals. Simulations and analysis show that HOT based LMS adaptive filter is computationally efficient and has fast convergence compared to LMS, NLMS and DFT-LMS. The computed Echo Return Loss Enhancement (ERLE), the general evaluation measure of echo cancellation, established the efficacy of proposed HOT based adaptive algorithm. In addition, the spectral flatness measure showed a significant improvement in cancelling the acoustic echo.

Monitoring Respiratory Activity Using PPG Signals by Order Reduced-Modified Covariance AR Technique

The clinical significance of certain cardiac arrhythmias can be understood only with reference to respiration. In normal healthy conditions, the respiratory rate is 10-20 breaths/minute. But, certain problems of illness, accidents or some other causes affect the regular sinus rhythm. A non-invasive, non-occlusive and non-intrusive respiration monitoring is desirable in a number of situations such as ambulatory monitoring, stress tests and sleep disorder investigations. Such methods are based on deriving the respiratory activity from the electrocardiogram (ECG). There have been several efforts on ECG-Derived Respiration (EDR). Presently research is being focused on PPG derived respiratory activity because of its simplicity in sensing the signal and proved a strong correlation between PPG and respiratory signals. In this paper, we present an efficient method for extraction of respiratory activity from photoplethysmographic (PPG) signals based on modified covariance method. The proposed method makes use of order reduced AR-model by restricting the pole locations in the frequency range of interest. Test results reveal that the order reduced-modified covariance AR model (OR-MCAR) has efficiently separated respiratory information from PPG than a normal AR model.

Dual-tree complex wavelet transform for motion artifact reduction of PPG signals

Ever since the medical device pulse oximeter was invented, reliable and accurate estimation of arterial blood oxygen saturation (SpO2), based on the differential absorption of red/infrared light by hemoglobins, has been a challenging task. The Photoplethysmogram (PPG) waveform, also known as the “pulse oximetry waveform”, is well recognized for its use in pulse oximetry applications for the estimation of SpO2 and can be obtained noninvasively and continuously in a comfortable manner using low cost & portable PPG sensors. Inaccuracy in the estimation of SpO2 may prevail due to the motion artifacts (MA) corruption in the detected PPG signals by the intentional or unintentional movements of a patient. The MA noise corruption is unavoidable while recording the PPGs because of a very small pulsatile component in PPG (0.1% of total signal amplitude) and it can be reduced by suitable processing of the PPG signals. In this paper, an approach for motion artifact (MA) reduction of photoplethysmographic (PPG) signals based on the concept of dual-tree complex wavelet transform technique is proposed. Experimental results revealed that DTCWT processing of MA corrupted PPGs outperformed the db10 wavelet processing for MA reduction of PPG signals and can be referred as best suitable MA reduction technique for pulse oximetry applications.

Extraction of respiratory activity from PPG and BP signals using Principal Component Analysis

In high risk situations such as cardiac arrhythmias, ambulatory monitoring, stress tests, sleep disorder investigations and post-operative hypoxemia situations, monitoring of respiratory activity would be mandatory. Electrocardiogram (ECG), blood pressure (BP) and photoplethysmographic (PPG) signals can be used for extraction of respiratory activity, and will eventually eliminate the use of additional respiratory sensor. Using a simple and standard non-parametric mathematical technique, Principal Component Analysis (PCA), the respiratory related information is extracted from complex data sets such as PPG and BP signals. The respiratory induced variations (RIV) of PPG and BP signals are described by coefficients of computed principal components. Singular value ratio (SVR) trend is used to find the periodicity, which is one of the crucial parameters in forming the data sets for PCA. Test results on MIMIC data base clearly indicated a strong correlation between the extracted and actual respiratory signals. Statistical measures in both time and frequency domains such as Relative Correlation Coefficient (RCC) and Magnitude Squared Coherence (MSC) respectively and Accuracy Rate (AR) are calculated to demonstrate the fact, that respiratory signal is present in the form of first principal components.

Estimation of respiratory rate from principal components of photoplethysmographic signals

Continuous monitoring of respiratory activity is mandatory in clinical, high risk situations such as ambulatory monitoring, intensive care, stress tests and sleep disorder investigations. Extraction of surrogate respiratory activity from electrocardiogram (ECG), blood pressure (BP) and photoplethysmographic (PPG) signals will potentially eliminate the use of additional sensor intended to record respiration. Principal Component Analysis (PCA) is a simple and standard non-parametric mathematical tool for extracting relevant information from complex data sets. In this paper, PCA is exploited for extraction of surrogate respiratory activity from PPG signals. The respiratory induced intensity variations (RIIV) of PPG signal are described by coefficients of computed principal components. Singular value ratio (SVR) trend is used to find the periodicity, which is one of the key parameters in forming the data sets for PCA. Test results on MIMIC data base clearly indicated a strong correlation between the extracted and actual respiratory signals. The evaluated similarity measures, both in time (RCC-Relative Correlation Coefficient) and frequency (MSC-Magnitude Squared Coherence) domains and calculated accuracy demonstrated the fact that respiratory signal is present in the form of first principal component.

On the performance of Time Varying Step-size Least Mean Squares(TVS-LMS) adaptive filter for MA reduction from PPG signals

Clinical Investigation of hypoxic status of the patients requires accurate information about the heart rate and oxygen saturation of arterial blood. Pulse oximeters are widely used for monitoring these parameters by recording the raw pulse oximeter signal, namely Photoplethysmogram (PPG). The recorded PPG Signal acquired using PPG sensors are usually corrupted with Motion Artifacts (MA) due to the voluntary or involuntary movements of patient. Reduction of MA has received much attention in the literature over recent years. In this paper, we present an efficient adaptive filtering technique based on Time Varying Step-size Least Mean Squares (TVS-LMS) algorithm for MA reduction. The novelty of the method lies in the fact that a synthetic noise reference signal for adaptive filtering, representing MA noise, is generated internally from the MA corrupted PPG signal itself instead of using any additional hardware such as accelerometer or source-detector pair for noise reference signal generation. Convergence analysis, SNR calculations and Statistical analysis revealed that the proposed TVS-LMS method has a clear edge over the Constant Step-size LMS (CS-LMS) based adaptive filtering technique. Test results, on the PPG data recorded with different MAs, demonstrated the efficacy of the proposed TVS-LMS algorithm in MA reduction and thus making it best suitable for real-time pulse oximetry applications.