Value of Magnetocardiography for the Non-Invasive Diagnosis of Coronary Artery Disease

Jan Lokies^a, Friederika Girke^a, Illya Chaikovsky^b, Michael Primin^b, Bernard Awolin^b,
Eckart Fleck^a

^aGerman Heart Center, Berlin, Germany
^bUniversity Witten/Herdecke, Essen, Germany

Correspondence: I Chaikovsky, University Witten/Herdecke, 45134 Waldsaum1 Essen, Germany.
E-mail: Illya_6@hotmail.com, phone +49 201 1059132, fax +49 201 1059140

1.  Introduction

Magnetocardiography (MCG) is a method for noninvasive registration and analysis of the electromagnetic field of the human heart. It has been suggested that analysis of the cardiac magnetic field may permit the non-invasive diagnosis of cardiac ischemia better than the 12-lead ECG. 

 Aim of this study was to find the way to represent electrical image of the heart obtained from magnetocardiographic data in comparison with perfusion image obtained from PET based on a few cases of diagnostically problematic patients with the absence of signs of ischemia on stress and rest ECG and unchanged function of left ventricle on echocardiogram.

2.  Material and Methods

110 consecutive patients (44% female and 56% male; mean age 62 +/- 10 years) with referred for evaluation of chest pain by elective invasive coronary angiography were prospectively studied by the MCG.

MCG was acquired from a four-channel SQUID-magnetometer. MCG recordings were taken from a total 36 measurement points within a 20 by 20 cm rectangular grid. Four recordings were taken at each of 9 positions. The sensor was positioned as close to the thorax as possible, with a standard reference position at the jugulum. Data were recorded at each position for 30 seconds with simultaneous registration of a lead II ECG and stored in a database for subsequent evaluation.

All patients underwent coronary angiography within 24 hours after the MCG examination. Coronary angiograms were acquired in multiple projections using the Judkins technique according to standard clinical practice.

All cardiac cycles were averaged for each measurement point using the R-peak of the lead II ECG as reference. Current density vector (CDV) maps were constructed based on the inverse problem solution ^that is implemented in the software package MagWin. The CDV maps were generated every 10 ms within the ST-T interval starting with the J-point up to the end of the T-wave. Each CDV map in course of the ST-T interval was analyzed visually by one independent experienced observer, who did not know the patients and results of the angiography. The visual analysis of CDV consisted of two steps: Step 1: classified each single map and Step 2: reassessed the entire set of maps of every patient based on the map results from step1.Criteria for classification have been the existence of a dipole structure, direction of main currents and presence or absence of an additional currents.

In addition, two variants (algorithm 1 and algorithm 2) of a fully automated computer analysis were tested. The first algorithm is based on the ratio of length of vectors directed left downwards (which is the normal direction for ventricular repolarization) to the length of vectors directed to other directions. This ratio was calculated for every map within the ST-T interval. If the average of ratios from all maps was contained within a defined interval, the computer result was rated either as MCG positive or negative. The second algorithm includes more than 50 parameters reflecting the inhomogeneity of ventricular repolarization [Chaikovsky at al, 2002].

Multivariate statistics are applied and an automatic classification rule was constructed. This rule represents Fisher's linear discriminate function (general view is D = b₁.x₁ + b₂.x₂ +...+b_nx_n + a where x1_, x_2, x_n- values of variables; b_1, b_2, b_n, a - coefficients; D - classified value of function) consisted of 5 variables, describing trend of magnetic field gradients between extrema of magnetic field in each map and interdependence of all 36 QRST curves within the measurement grid. Sensitivity, Specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) were calculated for the results by clinical observers and the two automated algorithms. These had been compared to the coronary angiography interpretations.

3.  Results

In 8 patients (7%), no diagnostic MCG-maps were obtained due to excessive external magnetic interferences. The prevalence of significant CAD in the remaining 102 patients was 49 %. 12 patients had one, 19 patients two, and 19 patients had three diseased vessels with more then 50 % lumen obstruction. In 6 patients, a vasospastic coronaropathy was diagnosed. The remaining 46 patients had either no coronary stenosis or less then with more than 50 % narrowing. The overall Sensitivity, Specificity, PPV and NPV of MCG for the detection of significant CAD are presented in Table1.

4.  Discussion

The data show that the MCG might be a useful tool to predict the presence of coronary artery disease and ischemia when analyzed by an experienced observer. Further, the results of a simple automatic decision system - algorithm 1 - were insufficient. However an algorithm 2 - clearly increased sensitivity. This demonstrates the need of better-defined criteria which also are suitable to be implemented in computerized algorithms.

Using advanced criteria sets, represented by algorithm II and especially by the experienced observer the MCG method has a fair diagnostic value when compared with qualitative coronary angiography. For the binary decision, presence or absence of coronary vascular obstruction, this totally noninvasive method shows great promise. Multicenter studies with larger numbers of patients should be conducted.

References

Chaikovsky I,Primin M,Nedayvoda I, Vasiliev V, Sosnitsky V, Steinberg F. Computerized classification of patients with coronary artery disease but normal or unspecifically changed ECG and healthy volunteers. In the proceedings of the 13-th International Conference on Biomagnetism, 2002, 411-414.