U. Leder

Mutual information and phase dependencies: measures of reduced nonlinear cardiorespiratory interactions after myocardial infarction

The heart rate variability (HRV) is related to several mechanisms of the complex autonomic functioning such as respiratory heart rate modulation and phase dependencies between heart beat cycles and breathing cycles. The underlying processes are basically nonlinear. In order to understand and quantitatively assess those physiological interactions an adequate coupling analysis is necessary. We hypothesized that nonlinear measures of HRV and cardiorespiratory interdependencies are superior to the standard HRV measures in classifying patients after acute myocardial infarction. We introduced mutual information measures which provide access to nonlinear interdependencies as counterpart to the classically linear correlation analysis. The nonlinear statistical autodependencies of HRV were quantified by auto mutual information, the respiratory heart rate modulation by cardiorespiratory cross mutual information, respectively. The phase interdependencies between heart beat cycles and breathing cycles were assessed basing on the histograms of the frequency ratios of the instantaneous heart beat and respiratory cycles. Furthermore, the relative duration of phase synchronized intervals was acquired. We investigated 39 patients after acute myocardial infarction versus 24 controls. The discrimination of these groups was improved by cardiorespiratory cross mutual information measures and phase interdependencies measures in comparison to the linear standard HRV measures. This result was statistically confirmed by means of logistic regression models of particular variable subsets and their receiver operating characteristics.

Postextrasystolic regulation patterns of blood pressure and heart rate in patients with idiopathic dilated cardiomyopathy

Assessment of fluctuations in heart rate (HR) following a premature ventricular complex (PVC) is valuable for identifying patients at high risk of sudden cardiac death. We hypothesised that postextrasystolic potentiation is the main determinant of the regulation patterns of blood pressure (BP) and HR following a PVC. Twelve patients with idiopathic dilated cardiomyopathy (IDC) and 13 control subjects with single PVCs (comparable coupling intervals) were investigated. Non‐invasive finger arterial BP and ECGs were analysed. Regulation patterns following a single PVC were quantified using the indices postextrasystolic amplitude potentiation (PEAP) and maximum turbulence slope of five consecutive mean BP values (MBP‐TS), and compared with the HR turbulence parameters turbulence slope (HR‐TS) and turbulence onset (HR‐TO). PEAP was significantly higher in IDC patients compared to controls (48.7 ± 32.6 vs. 9.8 ± 5.4 %, P < 0.01), whereas MBP‐TS was lower (0.97 ± 0.60 vs. 2.07 ± 1.04 mmHg BBI−1 (BBI, beat‐to‐beat interval), P < 0.05), as was HR‐TS (8.46 ± 7.90 vs. 30.73 ± 22.90 ms BBI−1, P < 0.01). HR‐TO was significantly higher in IDC patients (−0.56 ± 2.19 vs.−5.52 ± 4.13 %, P < 0.01). In addition, the regulation patterns of BP and HR following a single PVC differed significantly between IDC patients and controls. Specifically, we observed pronounced PEAPs in IDC patients. The baroreflex response initiated by the low pressure amplitude of the PVC was suppressed in IDC patients due to the augmented potentiation of the first postextrasystolic blood pressure. Furthermore, IDC patients displayed impressive postextrasystolic pulsus alternans phenomena, whereas healthy subjects exhibited a typical baroreflex pattern. The pulsus alternans phenomenon seems to be triggered by a PVC.

Non-invasive imaging of arrhythmogenic left-ventricular myocardium after infarction

A wide range of cardiac imaging procedures is available to assess left-ventricular function, geometry, perfusion, and altered metabolism. There is no non-invasive imaging procedure to detect electrically unstable myocardium. We assessed the electrical function of the myocardium by biomagnetic imaging of high-frequency depolarisation signals in a patient aged 70 years who had non-sustained ventricular tachycardia that developed after anterior left-ventricular myocardial infarction and apical aneurysm. Imaging was done on the basis of magnetic-field measurement and inverse computation of the magnitudes of 1200 left-ventricular endocardial dipoles regularly spaced (mean distance 3 mm). Such endocardial activity images have been computed for the unfiltered depolarisation signal (QRS) and for the last 40 ms of the bidirectional 30 Hz highpass filtered depolarisation signal (LP). In theory, three types of QRS/LP image pairs could appear. If areas of high dipole magnitude of LP and QRS images match, LP generators can be taken to lie in the intact myocardium. If areas of high dipole magnitude in LP images match areas of low dipole magnitude in QRS images, LP generators may belong to the area infarcted and, therefore, display properties of the surviving subendocardial layer. If no correlation of LP and QRS is seen, it will not fit into the concepts of arrhythmogenesis after infarction. By comparison of LP and QRS images (figure), we observed regional mismatch of dipole magnitudes. The QRS image shows low dipole magnitudes in the apical segment of the left ventricle (the apical aneurysm) with a loss of electrically active myocardium. The opposite has appeared in the LP image with high dipole magnitudes in the segments infarcted. This finding could be an indication of fragmented delayed intraventricular conduction in this region. We conclude that this method supports the imaging of delayed high-frequency depolarisation signals, which are known to correlate with conduction disturbances and, therefore, risk of arrhythmia. Our findings may help in assessing patients with malignant tachyarrhythmias after myocardial infarction.

Noninvasive biomagnetic imaging in coronary artery disease based on individual current density maps of the heart

OBJECTIVE In this paper we present an attempt at noninvasive imaging of distributed myocardial electrical activity in patients suffering from myocardial infarction and in healthy subjects. Although advances have been made, noninvasive three-dimensional imaging of cardiac electrophysiological activity is still in its infancy and extending our knowledge of cardiac electrophysiological properties may be a valuable guide in the treatment of patients with coronary artery disease. METHODS Magnetic field mapping data formed the input for an inverse solution that is based on a multiple dipole model. The lead field normalized minimum norm least square criterion was applied to predefined myocardial source geometry. Current density distributions were calculated for the left ventricle during ventricular depolarization. Images from two patients with previous myocardial infarction were compared to images from two healthy subjects. RESULTS Low regional and global current density was found in the infarction patients. Regions of low current density corresponded to infarcted segments. The images of the healthy subjects displayed less marked areas of low current density. CONCLUSION The proposed multiple dipole model may be able to distinguish viable from scarred myocardium. A prospective clinical study should be undertaken to investigate the spatial resolution and the diagnostic performance of this method.

Source localization in an inhomogeneous physical thorax phantom.

The influence of lung inhomogeneities on focal source localizations in electrocardiography (ECG) and magnetocardiography (MCG) is investigated. A realistically shaped physical thorax phantom with cylindrical lung inhomogeneities is used for electric and magnetic measurements. The lungs are modelled with a special ionic exchange membrane which allows different conductivity compartments without influencing the free ionic current flow. The dipolar current sources are composed of platinum wire and located at different depths and directions between the lung inhomogeneities. We localized the current dipoles with different boundary element method (BEM) models, based on electrical data and simultaneous electrical and magnetic data. Our results indicate the possibility of superadditive information gain by combining electrical and magnetic data for source reconstructions. We found a significant influence of the inhomogeneities on both the calculated source location and the calculated source strength. Mislocalizations of up to 16 mm and wrong dipole strengths of up to 52% were obtained when the lung inhomogeneities were not taken into account for source localization. Dipoles parallel to the lungs showed a larger localization error in depth than dipoles perpendicular to the lungs. We conclude that the incorporation of lung inhomogeneities will improve source localization accuracy in ECG and MCG.

Reproducibility of HTS-SQUID magnetocardiography in an unshielded clinical environment

A new technology has been developed which measures the magnetic field of the human heart (magnetocardiogram, MCG) by using high temperature superconducting (HTS) sensors. These sensors can be operated at the temperature of liquid nitrogen without electromagnetic shielding. We tested the reproducibility of HTS-MCG measurements in healthy volunteers. Unshielded HTS-MCG measurements were performed in 18 healthy volunteers in left precordial position in two separate sessions in a clinical environment. The heart cycles of 10 min were averaged, smoothed, the baselines were adjusted, and the data were standardized to the respective areas under the curves (AUC) of the absolute values of the QRST amplitudes. The QRS complexes and the ST-T intervals were used to assess the reproducibility of the two measurements. Ratios (R(QRS), R(STT)) were calculated by dividing the AUC of the first measurement by the ones of the second measurement. The linear correlation coefficients (CORR(QRS), CORR(STT)) of the time intervals of the two measurements were calculated, too. The HTS-MCG signal was completely concealed by the high noise level in the raw data. The averaging and smoothing algorithms unmasked the QRS complex and the ST segment. A high reproducibility was found for the QRS complex (R(QRS)=1.2+/-0.3, CORR(QRS)=0.96+/-0.06). Similarly to the shape of the ECG it was characterized by three bends, the Q, R, and S waves. In the ST-T interval, the reproducibility was considerably lower (R(STT)=0.9+/-0.2, CORR(STT)=0.66+/-0.28). In contrast to the shape of the ECG, a baseline deflection after the T wave which may belong to U wave activity was found in a number of volunteers. HTS-MCG devices can be operated in a clinical environment without shielding. Whereas the reproducibility was found to be high for the depolarization interval, it was considerably lower for the ST segment and for the T wave. Therefore, before clinically applying HTS-MCG systems to the detection of repolarization abnormalities in acute coronary syndromes, further technical development of the systems is necessary to improve the signal-to-noise ratio.

Kardiorespiratorische Desynchronisation nach akutem Myokardinfarkt

The prognosis of cardiac diseases can be estimated from the variability of regulation parameters of the cardiovascular system. Changes in the variability of a regulation parameter causes disturbances in the synchronisation of interacting control loops. Conclusions about the severity of the underlying functional impairment can be drawn from these disturbances. This study investigates the synchronisation of the control loops of the heart rate and respiration (cardiorespiratory synchronisation, CRS) after acute myocardial infarction.    We investigated 43 patients after myocardial infarction and 27 healthy controls. To quantify the CRS the synchronisation in phase of respiration and heart rate was assessed. The heart rate variability (HRV) was also assessed. Patients after myocardial infarction have a significantly reduced HRV and CRS. There is a non-linear relationship between HRV and CRS. Patients with left ventricular enlargement and reduced left ventricular ejection fraction (≤45%) significantly differed from the other infarct patients and controls in CRS but not in HRV. They had a marked degree of cardiorespiratory desynchronisation and were identified by a threshold value.    CRS is a measure of the interaction of respiration control and heart rate control. After myocardial infarction, a reduction of the HRV can be observed. The desynchronisation of the control loops of respiration and heart rate especially appears in large infarcts. This can be quantitatively assessed by the method presented. Die Variabilität von Regulationsgrößen des Herz-Kreislauf-Systems erlaubt Aussagen über die Prognose kardialer Erkrankungen. Die veränderte Variabilität einer Regulationsgröße führt zu Störungen in der Synchronisation interagierender Regelkreise. Die Quantifizierung dieser Störungen könnte Rückschlüsse auf die Schwere der zugrunde liegenden funktionellen Beeinträchtigung erlauben. Diese Studie untersucht die Synchronisation der Regelkreise von Herzfrequenz und Atmung (kardiorespiratorische Synchronisation, CRS) nach akutem Myokardinfarkt.    Es wurden 43 Patienten nach Myokardinfarkt und 27 Gesunde untersucht. Zur Quantifizierung der CRS wurde die Phasensynchronisation von Atemfrequenz und Herzfrequenz beurteilt. Ebenfalls wurden die Parameter der Herzfrequenzvariabilität (HRV) untersucht. Patienten nach Myokardinfarkt haben eine signifikant reduzierte HRV und CRS. Zwischen HRV und CRS besteht eine nichtlineare Abhängigkeit. Infarktpatienten mit linksventrikulärer Dilatation und eingeschränkter linksventrikulärer Pumpfunktion (EF≤45%) unterschieden sich in dieser Studie in der CRS, nicht aber in der HRV von den übrigen Infarktpatienten und Gesunden. Sie hatten eine ausgeprägte kardiorespiratorische Desynchronisation und konnten durch einen Schwellwert identifiziert werden.    Die CRS erfasst die Interaktionen von Herzfrequenz- und Atemregulation. Nach Myokardinfarkt kommt es zu einer Reduktion der HRV. Eine Desynchronisation der Regelkreise von Atmung und Herzfrequenz tritt offenbar insbesondere bei ausgedehnteren Myokardinfarkten ein. Dies kann mit der vorgestellten Methode quantifiziert werden.

Multichannel magnetocardiographic measurements with a physical thorax phantom

Artificial dipolar sources were applied inside a physical thorax phantom to experimentally investigate the accuracy obtainable for non-invasive magnetocardiographic equivalent current dipole localisation. For the measurements, the phantom was filled with saline solution of electrical conductivity 0.21 Sm−1. A multichannel cardiomagnetometer was employed to record the magnetic fields generated by seven dipolar sources at distances from 25 mm to 145 mm below the surface of the phantom. The inverse problem was solved using an equivalent current dipole in a homogeneous boundary element torso model. The dipole parameters were determined with a non-linear least squares fitting algorithm. The signal-to-noise ratio (SNR) and the goodness of fit of the calculated localisations were used in assessing the quality of the results. The dependence between the SNR and the goodness of fit was derived, and the results were found to correspond to the model. With SNR between 5 and 10, the average localisation error was found to be 9±8 mm, while for SNR between 30 and 40 and goodness of fit between 99.5% and 100%, the average error reduced to 3.2±0.3 mm. The SNR values obtained in this study were also compared with typical clinical values of SNR.

MCG simulations with a realistic heart-torso model

Magnetocardiograms (MCGs) simulated high-resolution heart-torso model of an adult subject were compared with measured MCGs acquired from the same individual. An exact match of the measured and simulated MCGs was not found due to the uncertainties in tissue conductivities and cardiac source positions. However, general features of the measured MCGs were reasonably represented by the simulated data for most, but not all of the channels. This suggests that the model accounts for the most important mechanisms underlying the genesis of MCGs and may be useful for cardiac magnetic field modeling under normal and diseased states. MCGs were simulated with a realistic finite-element heart-torso model constructed from segmented magnetic resonance images with 19 different tissue types identified. A finite-element model was developed from the segmented images. The model consists of 2.51 million brick-shaped elements and 2.58 million nodes, and has a voxel resolution of 1.56/spl times/1.56/spl times/3 mm. Current distributions inside the torso and the magnetic fields and MCGs at the gradiometer coil locations were computed. MCGs were measured with a Philips twin Dewar first-order gradiometer SQUID-system consisting of 31 channels in one tank and 19 channels in the other.

Passive vortex currents in magneto- and electrocardiography: comparison of magnetic and electric signal strengths

Vortex currents may be of importance in the early diagnosis of myocardial infarction caused by an occlusion of a coronary artery. We investigated the influence of a passive vortex current distribution, modelled by different conductivities in a hollow cylinder, on the localization error and on the signal strength in both the magnetocardiogram and the electrocardiogram. The hollow cylinder was mounted in a realistically shaped physical torso phantom. A platinum dipole was inserted into the cylinder. The compartment boundaries were modelled with two special ionic exchange membranes. The conductivity ratio of the cylinder compartment to the torso compartment was varied from 0.25 to 100. We compared the simultaneously measured magnetic and electric signal strengths as a function of this conductivity ratio. We found that an increasing conductivity ratio causes only a slight increase (about 19%) of the magnetic signal strength but a strong decrease (about 81%) of the electric signal strength. Using a homogeneous torso model, the dipole localization errors were, depending on the conductivity ratio, up to 16 mm. In conclusion, passive vortex currents might partially explain the differences between magnetocardiographic and electrocardiographic recordings observed both experimentally and clinically.