Avtor/Urednik		Korže, D; Zazula, D
Naslov		Izboljšan algoritem AZTEC za komprimiranje EKG signalov
Prevedeni naslov		Improved algorithm AZTEC for ECG signal compression
Tip		članek
Vir		In: Solina F, Zajc B, eds. Računalništvo in informatika, umetna inteligenca, razpoznavanje vzorcev, biomedicinska tehnika, robotika. 2. elektrotehniška in računalniška konferenca ERK'93: zbornik. Zvezek B. Ljubljana: Slovenska sekcija IEEE,
Leto izdaje		1993
Obseg		str. 359-62
Jezik		slo
Abstrakt		Paper presents the improved algorithm AZTEC, which fundamentals are given in (1). The essence of this method is, that the sequence of lines and slopes is stored instaed of the original signal. As it is well known, the ECG signal is composed of some more or less strait lines and some steeper segments (QRS complexes. It is shown, that the original ECG file can be compressed from 7 to 12 percent of its original size. After decompression of the compressed file the approzimation of the original ECG signal is received. It was noted, that the error in the area of the QRS complexes could be quite large (10 percent or more), particularly with the abnormal hearth beats. The large error was the reason for development of the improved AZTEC algorithm, where the bigger changes of the incline of the slope of signal was considered. The Q parameter is defined as the quotient of two successive slopes. In comparison with the normal AZTEC algorithm two or more slopes are generated instead of one. The error of the compression of the signal on areas with the QRS complexes with our improved algorithm becomes much smaller by factor 2 or 3. Our method can even be adequatly used for real-time ECG signal compression and for storing of the signal during the stress test. And even though the achived compression rates are not as with the original AZTEC algorithm, we can still compress the original ECG signal from 7 to 20 percent of his original size, depending on pulse and desired accuracy after the decompression. The influence of factors T, K and Q on the size of the compressed file and quality of the s ignal after decompression is studied. The quality of the decompressed signal is expressed with the average error on sample. It is desired that the absolute error of any sample is less then one percent of the maximum value of the signal, which also folds into the requested accuracy of the clinical ECG.
Deskriptorji		ELECTROCARDIOGRAPHY ALGORITHMS SIGNAL PROCESSING, COMPUTER-ASSISTED