Willis J. Tompkins

A Real-Time QRS Detection Algorithm

We have developed a real-time algorithm for detection of the QRS complexes of ECG signals. It reliably recognizes QRS complexes based upon digital analyses of slope, amplitude, and width. A special digital bandpass filter reduces false detections caused by the various types of interference present in ECG signals. This filtering permits use of low thresholds, thereby increasing detection sensitivity. The algorithm automatically adjusts thresholds and parameters periodically to adapt to such ECG changes as QRS morphology and heart rate. For the standard 24 h MIT/BIH arrhythmia database, this algorithm correctly detects 99.3 percent of the QRS complexes.

Quantitative Investigation of QRS Detection Rules Using the MIT/BIH Arrhythmia Database

We have investigated the quantitative effects of a number of common elements of QRS detection rules using the MIT/BIH arrhythmia database. A previously developed linear and nonlinear filtering scheme was used to provide input to the QRS detector decision section. We used the filtering to preprocess the database. This yielded a set of event vectors produced from QRS complexes and noise. After this preprocessing, we tested different decision rules on the event vectors. This step was carried out at processing speeds up to 100 times faster than real time. The role of the decision rule section is to discriminate the QRS events from the noise events. We started by optimizing a simple decision rule. Then we developed a progressively more complex decision process for QRS detection by adding new detection rules. We implemented and tested a final real-time QRS detection algorithm, using the optimized decision rule process. The resulting QRS detection algorithm has a sensitivity of 99.69 percent and positive predictivity of 99.77 percent when evaluated with the MIT/BIH arrhythmia database.

Electrotactile and vibrotactile displays for sensory substitution systems

Sensory substitution systems provide their users with environmental information through a human sensory channel (eye, ear, or skin) different from that normally used or with the information processed in some useful way. The authors review the methods used to present visual, auditory, and modified tactile information to the skin and discuss present and potential future applications of sensory substitution, including tactile vision substitution (TVS), tactile auditory substitution, and remote tactile sensing or feedback (teletouch). The relevant sensory physiology of the skin, including the mechanisms of normal touch and the mechanisms and sensations associated with electrical stimulation of the skin using surface electrodes (electrotactile, or electrocutaneous, stimulation), is reviewed. The information-processing ability of the tactile sense and its relevance to sensory substitution is briefly summarized. The limitations of current tactile display technologies are discussed, and areas requiring further research for sensory substitution systems to become more practical are suggested.<<ETX>>

ECG beat detection using filter banks

The authors have designed a multirate digital signal processing algorithm to detect heartbeats in the electrocardiogram (ECG). The algorithm incorporates a filter bank (FB) which decomposes the ECG into subbands with uniform frequency bandwidths. The FB-based algorithm enables independent time and frequency analysis to be performed on a signal. Features computed from a set of the subbands and a heuristic detection strategy are used to fuse decisions from multiple one-channel beat detection algorithms. The overall beat detection algorithm has a sensitivity of 99.59% and a positive predictivity of 99.56% against the MIT/BIH database. Furthermore this is a real-time algorithm since its beat detection latency is minimal. The FB-based beat detection algorithm also inherently lends itself to a computationally efficient structure since the detection logic operates at the subband rate. The FB-based structure is potentially useful for performing multiple ECG processing tasks using one set of preprocessing filters.

Comparing Reconstruction Algorithms for Electrical Impedance Tomography

An improved electrical impedance tomographic reconstruction algorithm is presented that is generally guaranteed to converge. The algorithm is attractive for several reasons. A modified Newton¿Raphson method varies a finite-element model of resistivities to fit a set of voltage measurements in a least-squared sense. Two procedures for calculating the Jacobian matrix are derived. One is standard, while the other is based on the compensation theorem. This second procedure is more efficient for computations, and just as accurate as the standard one. The inherent ill-conditioning in the approximate Hessian matrix of the linearized system is eliminated using the Marquardt method. Results from two-dimensional computer simulations are compared to four other reconstruction algorithms, which are based on methods proposed by other authors. The modified Newton¿Raphson method provided significantly better reconstructions than any of the other methods. The algorithms compared are the perturbation, equipotential, iterative-equipotential, and the double-constraint methods. The modified Newton¿Raphson method was found to be sensitive to measurement error, but future work in designing electrode-probing configurations is expected to reduce this sensitivity.

Estimation of QRS Complex Power Spectra for Design of a QRS Filter

We present power spectral analysis of ECG waveforms as well as isolated QRS complexes and episodes of noise and artifact. The power spectral analysis shows that the QRS complex could be separated from other interfering signals. A bandpass filter that maximizes the signal (QRS complex)-to-noise (T-waves, 60 Hz, EMG, etc.) ratio would be of use in many ECG monitoring instruments. We calculate the coherence function and, from that, the signal-to-noise ratio. Upon carrying out this analysis on experimentaly obtained ECG data, we observe that a bandpass filter with a center frequency of 17 Hz and a Q of 5 yields the best signal-to-noise ratio.

Neural-network-based adaptive matched filtering for QRS detection

The authors have developed an adaptive matched filtering algorithm based upon an artificial neural network (ANN) for QRS detection. They use an ANN adaptive whitening filter to model the lower frequencies of the electrocardiogram (ECG) which are inherently nonlinear and nonstationary. The residual signal which contains mostly higher frequency QRS complex energy is then passed through a linear matched filter to detect the location of the QRS complex. The authors developed an algorithm to adaptively update the matched filter template from the detected QRS complex in the ECG signal itself so that the template can be customized to an individual subject. This ANN whitening filter is very effective at removing the time-varying, nonlinear noise characteristic of ECG signals. The detection rate for a very noisy patient record in the MIT/BIH arrhythmia database is 99.5% with this approach, which compares favorably to the 97.5% obtained using a linear adaptive whitening filter and the 96.5% achieved with a bandpass filtering method.<<ETX>>

One-lead ECG for identity verification

This research investigates the feasibility of using the electrocardiogram (ECG) as a new biometric for human identity verification. It is well known that the shapes of the ECG waveforms of different persons are different but it is unclear whether such differences can be used to identify different individuals. In this research, we demonstrated successfully that it is possible to identify a specific person from a group of candidates using a one-lead ECG. A one-lead ECG, unlike two-dimensional biometrics, such as the fingerprint, is a one-dimensional, low-frequency signal that can be recorded from electrodes on the hands. This research applied two techniques, template matching and a decision-based neural network (DBNN), to implement the identity verification. Using each of the two methods separately on a predetermined group of 20 subjects, the experimental results showed that the rate of correct identity verification was 95% for template matching and 80% for the DBNN. Combining the two methods produced a 100% correct rate. Our results show that ECG analysis is a potentially applicable method for human identity verification.

A New Data-Reduction Algorithm for Real-Time ECG Analysis

Typically the ECG is sampled at a rate of 200 samples/s or more, producing a large amount of data that are difficult to store, analyze, and transmit. Data-reduction algorithms that operate in real time reduce he amount of data without losing the clinical information content. They must also leave sufficient computation time available for ECG analysis. We describe here a new algorithm called CORTES that is suited for such real-time applications. This algorithm combines the best features of two other techniques called TP and AZTEC. We present the results of a study to find optimal experimental values for the controlling variables in CORTES. We compare the computations of root-mean-square reconstruction errors for a diversity of encoded ECG signals.

Digital Filters for Real-Time ECG Signal Processing Using Microprocessors

Traditionally, analog circuits have been used for signal conditioning of electrocardiograms. As an alternative, algorithms implemented as programs on microprocessors can do similar filtering tasks. Also, digital filter algorithms can perform processes that are difficult or impossible using analog techniques. Presented here are a set of real-time digital filters each implemented as a subroutine. By calling these subroutines in an appropriate sequence, a user can cascade filters together to implement a desired filtering task on a single microprocessor. Included are an adaptive 60-Hz interference filter, two low-pass filters, a high-pass filter for eliminating dc offset in an ECG, an ECG data reduction algorithm, band-pass filters for use in QRS detection, and a derivative-based QRS detection algorithm. These filters achieve real-time speeds by requiring only integer arithmetic. They can be implemented on a diversity of available microprocessors.