Naslov
Real Time P-wave Detection in Esophageal ECG

Autori
Jeras, Miroslav ; Sovilj, Siniša ; Magjarević, Ratko

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
IFMBE Proceedings vol. 4 / Allen, Barry ; Lovell, Nigel - Sydney : World Congress on Medical Physics and Biomedical Engineering 2003, 2003

Skup
World Congress on Medical Physics and Biomedical Engineering 2003

Mjesto i datum
Sydney, Australija, 24.08.2003. - 29.08.2003

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
esophageal ECG; P-wave detection; wavelet transform

Sažetak
INTRODUCTION Transesophageal pacing is a method of temporary pacing which is implemented if the intracardial temporary lead cannot be implanted transvenously in emergency. Optimal position (implantation depth of the esophageal lead) is obtained through detection of the maximum of the magnitude of either P-wave (for atrial pacing) or the QRS complexes (for ventricular pacing). We have developed a method of real time P-wave detection in esophageal ECG (EECG). This method enables reliable detection of P-waves within a few tenths of millisecond, without a need on any information on a trailing QRS complex. METHODS The method used for signal decomposition is dyadic biorthogonal wavelet transform. As no search-back algorithms can be used in detection of P-waves in order to achieve fast response, the detection was achieved by a two-part algorithm. The first part is used for calculation of several statistical parameters of P-waves in the signal. The second algorithm uses the parameters from the first algorithm to search for waves that are similar to the average P-wave in last 10 beats. In both parts of the algorithm, several parameters are measured and/or calculated for each wave detected in the signal. The most important parameter in the analysis is the ratio between the amplitude of the 3rd and 4th scale of the wavelet transform of the current wave. Parameters are evaluated by a scoring function and the total score for each wave is compared to a threshold. The value of that threshold is determined experimentally on a known signal with annotated QRS complexes and P-waves. If the total score for a wave is greater than the threshold, the corresponding wave is detected as a P-wave. RESULTS The method was evaluated on EECG signals recorded on dogs using an implanted selective esophageal ECG lead. The performance of the detector was statistically evaluated and we obtained the algorithm accuracy 98, 1%, positive predictivity 99, 5% and sensitivity 98, 5%. DISCUSSION The difference of shapes and amplitudes of the P-wave of esophageal ECG signals measured at different lead implantation depths requires an adaptive algorithm for real-time P-wave detection. The adaptivity is achieved by using a more reliable search-back detection algorithm to determine properties of the current P-wave and by performing the real-time detection based on those parameters. CONCLUSION The presented real-time P-wave detection algorithm is based on the idea to use a separate non-real time search-back P-wave part and then to perform the real-time analysis based on statistical data of the collected P-waves. The detection algorithm was tested on esophageal ECG signals and showed high values of statistical performance measures.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika

POVEZANOST RADA