Author + information
- Illya Chaikovsky1,2,
- Georg Mjasnikov1,
- Michael Lutay3,
- Eugen Udovichenko4,
- Anton Popov4,
- Sergey Sofienko1 and
- Wenming Ji5
- 1Main Clinical Military Hospital of Ukraine, Kiev, Ukraine
- 2Institute of cybernetics, Kiev, Ukraine
- 3National Scientific Center “M.D. Strazhesko Institute of Cardiology, MAS of Ukraine, Kiev, Ukraine
- 4National Technical University “KPI”, Kiev, Ukraine, Ukraine
- 5University of Oxford, Oxford University Innovation, Oxford, United Kingdom
The diagnostic management of patients with chest pain remains a clinical challenge. In particular, existing non-invasive techniques may lack sensitivity and specificity to differentiate between epicardial and microvascular abnormalities. Magnetocardiography (MCG) is non-invasive evaluation of the magnetic field of the heart that is produced by the electric activity of the myocardium. Previous studies have shown that magnetocardiograms reveal obvious changes in patients with coronary artery disease (CAD) and normal electrocardiogram (ECG) at rest.
The objective of this research was the investigation of MCG value in differential diagnosis of coronary artery disease and coronary microvascular disease using novel approach of magnetocardiographic current density vectors (CDV) maps evaluation based on binary classification metric.
The study included 136 patients without a history of myocardial infarction. Coronary angiography was performed because of chest pain in all subjects. Depending on the results of coronary angiography, this group was divided into two subgroups: those with at least 50% stenosis in at least one of the main coronary arteries (subgroup 1, 82 subjects) and those without hemodynamically significant stenosis (subgroup 2, 54 subjects). All patients without hemodynamically significant stenosis underwent exercise ECG test and shown ST-segment depression during exercise testing. In all participants, the MCG examination was performed using a 9-channel MCG system located in an unshielded room.
The magnetocardiography recordings were taken from 36 positions at rest. From these CDV maps were generated during the ST-T interval. Each element of CDV map was described by two parameters: brightness, which corresponds to the current density in particular point and angle of current density vector at each point. As the result, 32 features were calculated for every map. Then binary k-NN classifier with various distance metrics (Cityblock, Mahalonobis, Chebychev, Eucledian) was used to qualify the just examined patient to the investigated categories.
The highest accuracy was demonstrated by binary k-NN classifier with Cityblock distance metric -124 (91%) patients were classified correctly. Sensitivity was 93%, specificity - 89%, positive predictive value -93% and negative predictive value –89%.
The MCG test at rest has the potential to be useful for the differential diagnosis between coronary artery disease and coronary microvascular disease.