Philip Langley

Principal Component Analysis as a Tool for Analyzing Beat-to-Beat Changes in ECG Features: Application to ECG-Derived Respiration

An algorithm for analyzing changes in ECG morphology based on principal component analysis (PCA) is presented and applied to the derivation of surrogate respiratory signals from single-lead ECGs. The respiratory-induced variability of ECG features, P waves, QRS complexes, and T waves are described by the PCA. We assessed which ECG features and which principal components yielded the best surrogate for the respiratory signal. Twenty subjects performed controlled breathing for 180 s at 4, 6, 8, 10, 12, and 14 breaths per minute and normal breathing. ECG and breathing signals were recorded. Respiration was derived from the ECG by three algorithms: the PCA-based algorithm and two established algorithms, based on RR intervals and QRS amplitudes. ECG-derived respiration was compared to the recorded breathing signal by magnitude squared coherence and cross-correlation. The top ranking algorithm for both coherence and correlation was the PCA algorithm applied to QRS complexes. Coherence and correlation were significantly larger for this algorithm than the RR algorithm(p < 0.05 and p < 0.0001, respectively) but were not significantly different from the amplitude algorithm. PCA provides a novel algorithm for analysis of both respiratory and nonrespiratory related beat-to-beat changes in different ECG features.

Frequency analysis of atrial fibrillation

Atrial fibrillation (AF) is a common arrhythmia which can have serious clinical consequences. Different characteristics of AF may be more amenable to treatment, but it is not easy to assess the characteristics of the atrial rhythm from ECG recordings since the atrial complexes are small relative to the ventricular complexes. AF is usually seen on the ECG as apparently irregular deviations of the baseline. Our aim was to develop a technique which would allow researchers to retrieve information about the atrial rhythm non-invasively from body-surface ECGs and to assess the stability of the atrial rhythm. We recorded 300 s of simultaneous 12-lead ECGs at 500 Hz directly to a computer from six patients with AF for subsequent analysis. For stability analysis, we split the 300 s recordings into two sections of 150 s each. These were then subdivided into sections of approximately 10 s, and principal component analysis was performed on each 12-lead section, generating 12 orthogonal data components. A frequency analysis of each component was carried out using a fast Fourier transform algorithm and the average spectrum calculated for each 150 s section. From this the dominant AF frequency was identified from the peak in the spectrum between 5 and 10 Hz. The atrial waveform was most commonly observed in principal components 6, 7 and 8 with mean (SD) frequency 6.8 Hz (0.9 Hz) and range 5.9 Hz to 8.2 Hz across the subjects. The difference between the paired 150 s sections was 0.3 Hz (0.3 Hz), showing that the dominant atrial frequency could be identified consistently and that there was little short term variability.

Comparison of atrial signal extraction algorithms in 12-lead ECGs with atrial fibrillation

Analysis of atrial rhythm is important in the treatment and management of patients with atrial fibrillation. Several algorithms exist for extracting the atrial signal from the electrocardiogram (ECG) in atrial fibrillation, but there are few reports on how well these techniques are able to recover the atrial signal. We assessed and compared three algorithms for extracting the atrial signal from the 12-lead ECG. The 12-lead ECGs of 30 patients in atrial fibrillation were analyzed. Atrial activity was extracted by three algorithms, Spatiotemporal QRST cancellation (STC), principal component analysis (PCA), and independent component analysis (ICA). The amplitude and frequency characteristics of the extracted atrial signals were compared between algorithms and against reference data. Mean (standard deviation) amplitude of QRST segments of V1 was 0.99 (0.54) mV, compared to 0.18 (0.11) mV (STC), 0.19 (0.13) mV (PCA), and 0.29 (0.22) mV (ICA). Hence, for all algorithms there were significant reductions in the amplitude of the ventricular activity compared with that in V1. Reference atrial signal amplitude in V1 was 0.18 (0.11) mV, compared to 0.17 (0.10) mV (STC), 0.12 (0.09) mV (PCA), and 0.18 (0.13) mV (ICA) in the extracted atrial signals. PCA tended to attenuate the atrial signal in these segments. There were no significant differences for any of the algorithms when comparing the amplitude of the reference atrial signal with that of the extracted atrial signals in segments in which ventricular activity had been removed. There were no significant differences between algorithms in the frequency characteristics of the extracted atrial signals. There were discrepancies in amplitude and frequency characteristics of the atrial signal in only a few cases resulting from notable residual ventricular activity for PCA and ICA algorithms. In conclusion, the extracted atrial signals from these algorithms exhibit very similar amplitude and frequency characteristics. Users of these algorithms should be observant of residual ventricular activities which can affect the analysis of the fibrillatory waveform in clinical practice.

Surface Atrial Frequency Analysis in Patients with Atrial Fibrillation: A Tool For Evaluating the Effects of Intervention

Introduction: The aims of this study were to evaluate (1) principal component analysis as a technique for extracting the atrial signal waveform from the standard 12‐lead ECG and (2) its ability to distinguish changes in atrial fibrillation (AF) frequency parameters over time and in response to pharmacologic manipulation using drugs with different effects on atrial electrophysiology.

Detection of sleep apnoea from frequency analysis of heart rate variability

Sleep apnoea is a clinical condition associated with a number of serious clinical and other problems. Patients who suffer from sleep apnoea have recurrent nocturnal apnoeas. The aim of this study was to assess the ability of an automated computer algorithm to detect sleep apnoea from the characteristic pattern of its recurrence, using RR interval data. Data from 35 training and 35 test subjects supplied by PhysioNet were analysed. To produce an algorithm which did not require highly accurate QRS detection, the QRS information supplied by PhysioNet were used without checking for artifactual data. Each subjects data were converted to a sequence of beat intervals, which was then analysed by Fourier transform. The study period varied from less than 7 hours to more than 10 hours. Patients with sleep apnoea tended to have a spectral peak lying between 0.01 and 0.05 cycles/beat, with the width of the peak indicating variability in the recurrence rate of the apnoea. In most subjects the frequency spectrum immediately below that containing the apnoea peak was relatively flat. The first visual analysis of the single computed spectrum from each subject led to a correct classification score of 28/30 (93%). The ratio of the content of the two spectral regions was obtained by dividing the area under the spectral curve between 0.01 and 0.05 cycles/beat by the area between 0.005 and 0.01 cycles/beat, and then a fixed threshold (3.15) was used to classify, the subjects automatically. The automated score for the training set was 27/30 (90%), 17/20 Apnoea (A), 10/10 Normal (C). The automated score for the test set was also 27/30 (90%).

Respiratory rate derived from principal component analysis of single lead electrocardiogram

We used principal component analysis to derive the respiratory rate from single lead ECGs. In this algorithm the respiratory induced beat-to-beat variability of the ECG is described by the coefficients of the principal components. Subjects were asked to breathe at different rates and naturally while respiration and ECG were recorded. Breathing rate, determined by Fourier analysis, was compared for the ECG derived respiration obtained by principal component analysis and the recorded respiratory signal. Across the different breathing patterns the mean absolute differences between reference respiratory rate and ECG respiratory rate were 0.2 breaths per minute (bpm) or less. In all cases the respiratory rate was accurately determined from respiratory surrogates derived from principal component analysis of the single lead ECGs.

Surface atrial frequency analysis in patients with atrial fibrillation: Assessing the effects of linear left atrial ablation

Introduction: Our group has shown previously that measurements of atrial frequency can be obtained from surface 12‐lead ECG recordings of patients during atrial fibrillation (AF), using a combination of principal component and Fourier transform algorithms. Such measurements are reproducible over time and change with drug manipulation of the arrhythmia.

A New Algorithm to Diagnose Atrial Ectopic Origin from Multi Lead ECG Systems - Insights from 3D Virtual Human Atria and Torso

Rapid atrial arrhythmias such as atrial fibrillation (AF) predispose to ventricular arrhythmias, sudden cardiac death and stroke. Identifying the origin of atrial ectopic activity from the electrocardiogram (ECG) can help to diagnose the early onset of AF in a cost-effective manner. The complex and rapid atrial electrical activity during AF makes it difficult to obtain detailed information on atrial activation using the standard 12-lead ECG alone. Compared to conventional 12-lead ECG, more detailed ECG lead configurations may provide further information about spatio-temporal dynamics of the body surface potential (BSP) during atrial excitation. We apply a recently developed 3D human atrial model to simulate electrical activity during normal sinus rhythm and ectopic pacing. The atrial model is placed into a newly developed torso model which considers the presence of the lungs, liver and spinal cord. A boundary element method is used to compute the BSP resulting from atrial excitation. Elements of the torso mesh corresponding to the locations of the placement of the electrodes in the standard 12-lead and a more detailed 64-lead ECG configuration were selected. The ectopic focal activity was simulated at various origins across all the different regions of the atria. Simulated BSP maps during normal atrial excitation (i.e. sinoatrial node excitation) were compared to those observed experimentally (obtained from the 64-lead ECG system), showing a strong agreement between the evolution in time of the simulated and experimental data in the P-wave morphology of the ECG and dipole evolution. An algorithm to obtain the location of the stimulus from a 64-lead ECG system was developed. The algorithm presented had a success rate of 93%, meaning that it correctly identified the origin of atrial focus in 75/80 simulations, and involved a general approach relevant to any multi-lead ECG system. This represents a significant improvement over previously developed algorithms.

An algorithm to distinguish ischaemic and non-ischaemic ST changes in the Holter ECG

Changes in the ECG ST segment are often observed in patients with myocardial ischaemia. However, non-ischaemic changes in ST level are also common thereby limiting ischaemia detection accuracy. The aim of this study was to devise an algorithm and determine its accuracy in distinguishing between ischaemic and non-ischaemic changes in the ECG ST-segment, using expertly annotated ECG data sets as a gold standard reference. The algorithm considered only the change in ST relative to a baseline ST level (/spl Delta/ST) provided by the PhysioNet database, and based on simple level thresholding within specified time windows. An initial score of 82.3% (accuracy 91.1%, with sensitivity 99.0% and specificity 88.8%) was achieved for the learning set. By making slight modifications to the algorithm and introducing principal components of ST it was not possible to improve the original algorithm. The original algorithm was therefore left as our challenge entry achieving an accuracy of 90.7% for the test data set (score of 81.4%, entry 1, 1 May 2003).

Effect of changes in heart rate and in action potential duration on the electrocardiogram T wave shape

The mechanisms responsible for changes in T wave symmetry and amplitude with changes in heart rate and action potential duration were investigated. A computer model of normal left ventricular repolarization was used to simulate the T waves on the surface 12-lead ECG. The effect of heart rate changes was studied by varying the ratio between dispersion of repolarization (Disp) and action potential repolarization duration (APRD). With constant dispersion. as heart rate increases, APRD decreases and the ratio Disp/APRD increases. T waves were simulated while varying the Disp/APRD ratio from 3.6% to 100%. The T wave symmetry ratio measured from the areas either side of the peak (SRarea), the symmetry ratio from the times either side of the peak (SRtime) and the T wave amplitude (Tamplitude) were calculated from each simulated ECG. SRarea decreased from 1.42 to 0.77, SRtime from 1.75 to 1.04 and the Tamplitude increased from 0. 19 mV to 2.30 mV. The stability of results with variation in model characteristics was also investigated, by moving the heart +/- 20 mm on all three axes, rotating the heart axes by +/- 10 and by modifying all constants defining the action potential by +/- 5% and +/- 10%. T wave amplitude was sensitive to changes in heart position, as the heart was moved towards the body surface. However, T wave shape changed very little with heart position or rotation, with the SD of SRarea varying by less than 0.05 over an SRarea range of 0.65 for different values of Disp/APRD ratio. We have shown from our model that cardiac T waves increase in amplitude, and become more symmetric with their peaks becoming central as APRD shortens with increasing heart rate, agreeing with clinical observations. These results help to explain the T wave shape changes which occur when heart rate increases.