Preliminary results in display and abnormality recognition of Hilbert transformed e.c.g.s

An application of the Discrete Hilbert Transform (d.h.t.) in the area of e.c.g. display and analysis is presented. The implementation of the Hilbert Transform is discussed and a comparison of the resulting display is made with that obtained when displaying vectorcardiograms. Other features of d.h.t. e.c.g.s are also presented and in this instance two techniques for pattern recognition of abnormalities are mentioned. The first deals with overall features of the individual e.c.g. beats and is probability based. The second technique is more concerned with the beat to beat variations in the e.c.g. and is more syntactic in nature. A comparison of the techniques is given along with computer generated results.     

Computer methods for analysing the high-frequency electrocardiogram

The high-frequency electrocardiogram (e.c.g.) is similar to the clinical e.c.g. commonly interpreted by clinicians except that a wider recording bandwidth is used. This extended signal bandwidth permits more information about the cardiac electrical activity to be perceived than can be seen in a clinical e.c.g. The high-frequency data appear as waveform features called notches and slurs. The notches have been correlated with coronary heart disease by several investigators. Most of the work in high-frequency electrocardiography has been done by manual techniques with minimal automation of the signal analysis. The purpose of this project was to develop computer algorithms for the automatic recognition and characterisation of notch parameters. Investigations in this study were conducted in both the time and frequency domains using digital filtering and fast Fourier transform techniques. Methods for notch recognition and quantification by computers were evaluated.     

Effect of electrode parameters on the bandwidth of the surface e.m.g. power-density spectrum

The effect of the electrode configuration (unipolar against bipolar), the electrode separation, the electrode position and the electrode contact area on the bandwidth of the power density spectrum of the surface e.m.g. has been studied. The dependence of the half-power point bandwidthB on the interelectrode distanced can be described by the regression function B_Hz=810/d_mm+58. The unipolar electrode configuration yields an e.m.g. signal with the smallest bandwidth. Neither a variation of the electrode position nor a variation of the electrode contact area has a significant effect on the bandwidth of the e.m.g.     

Summation of fibre potentials and the e.m.g.-force relationship during the voluntary movement of a forearm

A linear mathematical model of the electromyogram (e.m.g.) has been developed for the biceps muscle. The number of motor units (and therefore muscle fibres) contributing to the resultant e.m.g. at any stage of movement has been found from the force analysis of elbow flexion. The depths of various motor units and the phase difference between the recruitment of any two motor units have been formulated using a spiral spread of recruitment sequence. The attenuation of individual motor-unit action potentials due to varying depths has been taken into consideration, and due regard has been taken of the length-tension diagram of a muscle while performing the force analysis. Attention has been focused on the flexion of the elbow joint, in which a method of finding the individual contribution of the biceps and brachialis muscles has been developed and applied. The results predicted by the model have been verified by experiments. The model can also be extended to the e.m.g. of other fast skeletal muscles. The conditions and limitations for such generalisations have been stated and discussed.     

Cerebral potentials and leg muscle e.m.g. responses associated with stance perturbation

Summary In order to investigate the neuronal mechanisms underlying the compensatory movements following stance disturbance, leg muscle e.m.g. responses and cerebral potentials evoked by a treadmill acceleration impulse were analysed. It was found that the displacement was followed by a cerebral potential of a latency of 40–45 ms and EMG responses in the calf muscles at a latency of 65–70 ms. The e.m.g. responses represented specific compensatory reactions to the mode of perturbation (with a gastrocnemius activation following positive acceleration but a tibialis ant. activation following negative acceleration). The cerebral potentials, however, showed a common pattern to both conditions. In addition, the leg muscle e.m.g. reactions were not altered by learning effects and by forewarning of displacement onset, while the amplitude of the cerebral potentials was significantly smaller in these conditions compared to those produced in response to randomly induced perturbations. It was therefore concluded that the leg muscle e.m.g. reactions are mediated by a polysynaptic spinal reflex pathway which depends on a supraspinal control. The cerebral potentials seem to represent afferent signals which can be supposed to be subjected to modification and processing by supraspinal motor centres, according to the actual requirements.This work was supported by the Deutsche Forschungsgemein-schaft SFB 70 — Hirnforschung und Sinnesphysiologie — and Be 936/1-1     

Changes of the e.e.g. during a c.n.v. paradigm

An experimental set-up and a method of analysis are described which permit the performance of e.e.g. analysis under contingent negative variation (c.n.v.) conditions with repeated stimulation. Investigations of six subjects showed a significant reduction in the average power in the alpha band immediately after the imperative stimulus, which lasted only about 1 second. In the interstimulus interval during the c.n.v. a less marked power reduction could be observed. Since the reduction of the average power in the alpha band was approximately the same for both hemispheres and since it was also dependent on the modality of the stimulus, an interrelation-ship is assumed between the directed attention and the reduction observed. This appears to allow the possibility of objectively analysing cortical functions by means of an analysis of scalp recordings under c.n.v. conditions.     

Design and evaluation of QRS and noise detectors for ambulatory e.c.g. monitors

In ambulatory e.c.g. monitoring, a high level of noise results in false QRS detection. We present a compact low-powered QRS detector design. We present a noise detector circuit that identifies most false QRS detections so that it can disable alarms. We present a standardised exercise protocol for testing the QRS detector. Data collected from subjects bending forward (for baseline drift), lifting a weight (for e.m.g.), and jogging (for motion artefact) present a realistic test set for an ambulatory QRS detector. We observe error rates of the order of 1%, the noise detector identifies more than half of these. These techniques should reduce false alarms in arrhythmia monitoring systems.     

Model of the heart and its applications to the e.c.g.

The human body with the heart as an active electrical source has been modelled by three sets of harmonic polyphase generator voltages or spinning vectors; each set being confined to one of the three standard planes, i.e. the frontal plane, the horizontal plane and the sagittal plane. The equivalent circuit for one of the standard planes, the frontal plane, has been developed and analysed. The identical method of approach for the analysis of e.c.g. voltages in the two other planes has been pointed out and a typical multidimensional case analysis of this model is given. This model has been shown to satisfy a general version of the Burger law and also a general version of the Goldberger augmented lead voltages. The modified Wilson central terminal (w.c.t.) has also been found not to be at the electrical zero of the system. it has further been observed, and confirmed by experiment, that the vector sum of harmonic currents towards the modified w.c.t. is zero, irrespective of the different values of externally connected resistors and without assuming the modified w.c.t. at the electrical zero of the system, i.e. Kirchhoff’s harmonic-current law is satisfied at the modified w.c.t. For equal values of external resistors, the vector sum of the phase voltages and augmented lead voltages also become zero. The model has further been used to study the effect of a number of standard leads placed in the frontal plane while recording the e.c.g., on the maximisation of: (a) the e.c.g. voltages between the standard leads placed over the human body; (b) the e.g.c. voltage between the modified w.c.t. and a reference point on the body; (c) the ratio of the e.c.g. Goldberger voltage to the corresponding phase voltage; and (d) the values of the e.c.g. phase voltages. The effects of the number of electrodes placed in the horizontal and sagittal planes on the respective groups of e.c.g. voltages are discussed. By controlling the number of electrodes in any one of the three standard planes, when the e.c.g. voltages are maximised, it has been shown that the signal-to-noise ratio for those e.c.g. voltages will be increased.     

The USH2A c.2299delG mutation: dating its common origin in a Southern European population

Usher syndrome type II is the most common form of Usher syndrome. USH2A is the main responsible gene of the three known to be disease causing. It encodes two isoforms of the protein usherin. This protein is part of an interactome that has an essential role in the development and function of inner ear hair cells and photoreceptors. The gene contains 72 exons spanning over a region of 800 kb. Although numerous mutations have been described, the c.2299delG mutation is the most prevalent in several populations. Its ancestral origin was previously suggested after the identification of a common core haplotype restricted to 250 kb in the 5′ region that encodes the short usherin isoform. By extending the haplotype analysis over the 800 kb region of the USH2A gene with a total of 14 intragenic single nucleotide polymorphisms, we have been able to define 10 different c.2299delG haplotypes, showing high variability but preserving the previously described core haplotype. An exhaustive c.2299delG/control haplotype study suggests that the major source of variability in the USH2A gene is recombination. Furthermore, we have evidenced twice the amount of recombination hotspots located in the 500 kb region that covers the 3′ end of the gene, explaining the higher variability observed in this region when compared with the 250 kb of the 5′ region. Our data confirm the common ancestral origin of the c.2299delG mutation.     

Online analysis of e.c.g. arrhythmias with a digital computer

Our online program was developed to identify premature atrial and ventricular contractions and missed beats. Special care is taken with the recognition of artefacts. 14 patients with a high rate of arrhythmias were studied and 108 000 beats analysed independently by cardiologists and correlated with the computer analysis. Of a total of 7458 premature ventricular contractions, 411 events were detected false positives or negatives. This is an error of 5%. Out of the 826 premature atrial contractions 140 mistakes were found (17%). In a group 1772 missed beats (av-block etc.) 159 were improperly identified (9%). Of 1410 ventricular ectopics without prematurity there were 164 incorrect results (12%). The vast majority of muscle potentials and artefacts was properly classified and detected (error 11–15%). The main reason for the errors is the bad signal-to-noise ratio experienced from time to time. With the help of a digital computer a reliable analysis of the most important arrhythmias in intensive care patients has been achieved. A main part of the program should be concerned with artefact detection.