Differences between normal and arrhythmic electrocardiograms using frequency analysis

Abstract

We report an advance of a research project basedon the hypothesis that electrocardiograms (ECG)are easier to interpret in the frequency domain than in the time domain. This paper is aimed tomedical personnel. Methodology: 21 electrocardiograms(eighteen normal and three arrhythmic)were taken from a database of MassachusettsInstitute of Technology, and analyzed in thefrequency domain, in order to distinguishquantitative and qualitatively spectra ofarrhythmic patients from spectra of healthypatients. We worked on a small number of sampleslooking for stationary conditions, calculating thepower spectral density (PSD) on intervals of 300samples, which enclose two or three cardiac cycles,over a total range of 90.000 samples. Forqualitative identification, the correspondent draftswere etched and their morphology compared. Forquantitative identification, an average of the PSDof healthy registries we made, to which a distancewere measured (how much an arrhythmic signaldiffers from the average of healthy signals) anddifferences of its integral to each arrhythmic PSD.Results: We found, preliminarily, differencesbetween healthy and non healthy signals. Thepaper presents the drafts, distances and integralsof PSD of studied registries. Conclusion: In thefrequency domain is possible to establishmorphologic and numeric differences, easy toidentify, between healthy and non healthyelectrocardiograms.

Keywords

• Physiology
• electrocardiogra
• signal processing
• computer-assisted