Comparison of a new template matching algorithm, correlation, and area of difference methods for detection of ventricular tachycardia

Template matching algorithms have been shown to discriminate intraventricular electrograms of atrial and ventricular origin and may prove useful in implantable cardioverter defibrillators (ICDs). Those methods which normalize for amplitude and differentiate well require multiplications which are costly to ICD battery charge. Signature analysis (SIG) was developed which normalizes for amplitude and does not require multiplications. SIG was compared to correlation waveform analysis (CWA), amplitude normalized area of difference (NAD), and unnormalized area of difference (AD) for discrimination of ventricular tachycardia (VT) from sinus tachycardia (ST). After calculating prospective thresholds, the average percent depolarizations misclassified for ST and VT records, with/without baseline normalization (BN) were: CWA 9%/23%, NAD 11%/27%, SIG 13%/22%, and AD 12%/25%. Therefore, BN improved results markedly and, due to similar results, SIG or AD is preferable because of reduced computation