Electrocardiogram (ECG) compression techniques are compared, and a unified view of these techniques is established. ECG data compression schemes are presented in two major groups: direct data compression and transformation methods. The direct data compression techniques are ECG differential pulse code modulation (DPCM) and entropy coding, AZTEC, Turning-point, CORTES, Fan and SAPA algorithms, peak-picking, and cycle-to-cycle compression methods. The transformation methods include Fourier, Walsh, and Karhunen-Loeve transforms. The theoretical bases behind the direct ECG data compression schemes are presented and classified into three categories: tolerance-comparison compression, DPCM, and entropy coding methods. A framework for evaluation and comparison of ECG compression schemes is presented. >

ECG data compression techniques-a unified approach

An adaptive method for performing optimal time-varying filtering of nonstationary signals having a recurring statistical character, e.g., recurring pulses in noise, has been proposed. This method, called time-sequenced adaptive filtering, is applied to the enhancement of abdominally derived fetal electrocardiograms against background muscle noise. It is shown that substantial improvement in terms of signal distortion is obtained when time-sequenced filtering, rather than conventional time-invariant filtering, is employed. The method requires two or more abdominal channels containing correlated signal components, but uncorrelated muscle noise components. The location of the fetal pulses in time must be estimated in order to synchronize the filter's time-varying impulse response to the fetal cardiac cycle.

/pdf/fetal-electrocardiogram-enhancement-by-time-sequenced-4gug1m1wee.pdf

Fetal Electrocardiogram Enhancement by Time-Sequenced Adaptive Filtering

A new algorithm for QRS delineation has been developed. Based on the envelope of the e.c.g. signal a delineation function is defined, which yields a single positive pulse for each complex. From this function the onset and end of the QRS or, alternatively, a fiducial point is determined. To remove low-frequency component such as S-T abnormalities without distortion of the QRS complex, a filter with time-varying characteristics is used. The accuracy of the method has been evaluated in a test set of different QRS complexes obtained from coronary care patients. For QRS onset, the standard deviation of the difference between automated and manual determination was 7 ms in normal beats and 14 ms in ectopic beats. With simulated noise added to each waveform an average dispersion of 7 ms was observed in the recognition of the QRS onset at a signal-to-noise ratio of 15 dB. The corresponding dispersion in the location of a fiducial point was 2 ms. Using simulated e.c.g. data, the stability of the method is demonstrated for transitions between different waveform morphologies.

Delineation of the QRS complex using the envelope of the e.c.g.

A preliminary investigation of a wavelet packet based algorithm for the compression of single lead ECG is presented. The algorithm combines the efficiency and flexibility of wavelet packet expansions with the methodology of the Karhunen-Loeve transform (KLT). For selected records from the MIT-BIH arrhythmia database, an average data rate of 184.7 bits per second, corresponding to a compression ratio of 21.4:1, is achieved. When compared with the KLT applied to the same data, the wavelet packet algorithm generates significantly lower data rates with less than one-third the computational effort.

Wavelet packet-based compression of single lead ECG

Signal processing of electrocardiographic signals has a long and rich history and has greatly helped to enhance the diagnostic capability, especially when signals are recorded in noisy environments. This article gives an introductory presentation of the basic set of algorithms used for conditioning the ECG with respect to different types of noise and artifacts, detecting heartbeats, extracting basic ECG measurements, and performing data compression. Signal processing in clinical applications is exemplified by the high-resolution ECG and T wave alternans.

/pdf/electrocardiogram-ecg-signal-processing-ij4ngbqv7v.pdf

Electrocardiogram (ECG) Signal Processing

The problem of filtering artifact from QRST complexes of ECG records has been approached by the use of a digital, time-varying filter The optimum form of this filter assuming uncorrelated, zero-mean, additive noise is presented and shown to be equivalent to estimating the coefficients of an intrinsic component expansion of the complex Results are reported showing that QRST complexes in limb leads of normals may be accurately represented by only eight terms of the intrinsic component expansion A suboptimum filter is presented and results reported which show that an order of magnitude reduction of the 005 to 100 Hz artifact energy in the filtered complex can be achieved

The Use of Intrinsic Components in an ECG Filter

Ventricular fibrillation, a form of electrical instability of the myocardium, is strongly suspected to be a significant contributor to ``sudden'' coronary fatalities. Present evidence indicates that ventricular fibrillation is often preceded by other forms of abnormal electrical activity of the heart, especially ventricular premature beats (VPBs). Early detection of the presence of VPBs in the electrocardiogram may be important in making medical decisions to avert a later coronary attack.

Phase Detection of R Waves in Noisy Electrocardiograms

A linearized model for the transmission of a plane wave through a single component, two-phase bubbly mixture has been developed. The analysis starts with an investigation into the compressibility of a single bubble under the influence of a pressure pulse. These data are then used in a scattering model to predict speed of sound. Results were obtained for speed of sound as a function of frequency. These results show clearly the effects of phase change and heat transfer at low frequencies with resonance and cut-off at higher frequencies. The experimental model achieved a measurement of the speed of sound by passing an acoustic tone burst through the medium and measuring transit time—either directly or as a phase lag. The mixture was created by passing saturated vapor through a porous metal filter and allowing the bubbles to rise in a saturated liquid. An attempt (only partially successful) to measure vapor fraction at the level of the acoustic transducers with a differential pressure measurement was made. Experiments were carried out with both H2O and Refrigerant-12. Qualitative verification of this was possible with some quantitative discrepancies in the resonant frequency.

Charles L. Feldman

Papers

The Use of Intrinsic Components in an ECG Filter

Phase Detection of R Waves in Noisy Electrocardiograms

The speed of sound in single-component two-phase fluids: theoretical and experimental†

The Use ofIntrinsic Components inan ECG Filter