Adaptive filtering in exercise high resolution ECG as applied to the hypertrophic cardiomyopathy

The application of adaptive filtering to ECG signals has been investigated for many years. This study shows that the exercise high resolution ECG (HRECG) can also be processed successfully in a similar way. Two groups were included consisting of 20 healthy individuals and 24 patients with hypertrophic cardiomyopathy (HCM). The HRECG parameters for both groups were similar (QRSdur: 107 +/- 7 vs 114 +/- 18 ms NS, LAS: 25 +/- 8 vs 22 +/- 6 ms NS). In the first step, the HRECG signal was acquired at rest to obtain the averaged reference pattern. The next step was associated with peak exercise in which one could calculate short duration averaging (approximately 30 beats) or apply adaptive filtering in which the exercise component (EC) was extracted. Exercise was performed in the supine position on a bicycle ergometer. The load of 50 W was incremented by 50-W steps in 3-minute intervals and the test was ended by fatigue. Signals were recorded in X, Y, and Z bipolar leads with a 20-Hz high pass filter. The short time average QRS duration mostly was abbreviated in normal individuals in contrast to HCM patients in which ventricular activity prolonged with sensitivity, specificity, and negative and positive predictive values: 79%, 65%, 73%, and 72%, respectively. The adaptive recurrent filtration (ARF) after cutoff of the EC at the level of 70 ms (this level is the EC mean value of both groups) showed the following statistics: 63%, 90%, 88%, and 90%. The Student's t-test as applied to the duration of EC allowed a statistically significant difference between normals and HCM patients (66 +/- 4 vs 71 +/- 6 ms, P < 0.0052) and between HCM patients with and without ventricular tachyarrhythmia and DS (74 +/- 6 vs 69 +/- 6 ms, P < 0.046).