Effects of filtering on computer-aided analysis for detection of chronic ischemic heart disease with unshielded rest magnetocardiography mapping

Recent studies have reported better sensitivity of magnetocardiographic (MCG) mapping, as compared to ECG, in detecting ventricular repolarization (VR) abnormalities due to myocardial ischemia in patients (pts) with Ischemic Heart Disease (IHD). For quick data reduction, automatic analysis of MCG mapping is mostly used. The aim of our study was to evaluate if filtering modality could alter automatic analysis of MCG. Method: 39 subjects were studied: 20 normals and 19 IHD pts, with angiography-documented >70% coronary stenosis, positive stress/SPECT and ischemic 12-lead ECG in 12/19 (63%). Rest MCG was recorded with a 36-channel system (at 1 kHz; bandwidth DC-100 Hz). To assess VR, Hänninen's STα angle and three magnetic field dynamics parameters, [i.e. +/- poles: angle (A), distance (D) and ratio (R)] during the T-wave interval, were computed from the same MCG maps: 1) after digital 20 Hz low-pass filtering (LPF) and 2) after digital 50 Hz adaptive filtering (AF). The baseline was unchanged. Three quantitative MCG scores of the T-wave (EXT, ML, Q) were automatically calculated (with 20 Hz LPF only). Results: Whereas the filtering modality didn't affect the predictivity of the STα angle, the predictive values of A, D, and R were different and partially contradicting. Automatic MCG scores had a predictive values ranging between 73% and 92%. Conclusions: The diagnostic power of unshielded MCG for detection of chronic IHD, with T-wave parameters (A, D and R) might be affected by LPF. The STα angle is not affected by LPF. Automatic EXT, ML and Q scores have better predictivity than ECG.