## As

vs vs vs

Figure 2.9. PR Logic algorithm as developed by Medtronic. The two cycles before the event are broken down into segments so that the result can be compared with a library of arrhythmias and a decision can be taken whether the arrhythmia is of ventricular origin.

AV relation are used for identification of atrial tachyarrhythmias. Stability of RR intervals and AV dissociation are used to identify ventricular arrhythmias when atrial fibrillation is present [22].

In both algorithms therapy is delivered unless a discriminator identifies an atrial tachyarrhythmia.

### Dual-chamber Algorithms Based on Rate Branches

Comparison of atrial and ventricular rates is applied in several algorithms. The dual-chamber algorithms in Biotronik (Figure 2.10) and St. Jude Medical initially divide tachyarrhythmias into three rate branches: ventricular rate > atrial rate, ventricular rate < atrial rate, and ventricular rate = atrial rate. In the latter two branches, applicable single- and dual-chamber arrhythmia discriminators are applied to classifiy the arrhythmia [23]. In case of the ventricular rate = atrial rate branch, the onset criterion and analysis of the

Figure 2.10. SMART algorithm (Biotronik), with three rate branches. PP: atrial intervals; RR: ventricular intervals; SVT: supraventricular tachycardia; VT: ventricular tachycardia.
Figure 2.11. Atrial view (algorithm developed by Guidant). It compares atrial with ventricular rate once the threshold for ventricular tachycardia (VT) is reached. Four zones are defined (bradycardia, VT, fast VT, and ventricular fibrillation).

atrioventricular relationship are applied. The association or dissociation of the rhythms is monitored based on the stability criterion. If the ventricular rhythm is stable and the atrial rhythm is unstable, the tachyarrhythmia will be classified as ventricular. If both rhythms are stable, the stability of the atrioventricular relationship is analysed to exclude atrioventricular dissociation. In Guidant dual-chamber devices (Figure 2.11), priority is given to the single-chamber detection criteria onset and stability. An aggressive programming of single-chamber detection criteria in these devices will decrease the sensitivity but increase the specificity of arrhythmia discrimination. The dual-chamber detection criterion 'ventricular rate > atrial rate' can be applied to prevent underdetection of ventricular arrhythmias. The 'Afib threshold' criterion cannot prevent inappropriate arrhythmia classification as priority is given to the stability criterion.

### Performance of Dual-chamber Algorithms

The majority of studies conducted with dual-chamber ICDs were restricted to one manufacturer [21-24]. These studies mainly focused on the feasibility and safety of dual-chamber devices, and provided data for improved specificity of arrhythmia detection without compromising the sensitivity for ventricular tachyarrhythmias. The specificity ranged between 66.7 and 93.3% with positive predictive values for ventricular tachyarrhythmias between 87.4 and 98.4%. These data support an actual benefit of dual-chamber devices over single-chamber devices.

The Atrioverter: Detection of Atrial Tachyarrhythmias 17

Randomized studies comparing single-chamber and dual-chamber ICDs have been performed [25-29]. However, conclusive evidence of the superiority of dual-chamber over single-chamber discrimination has not been proven [30]. The functionality of dual-chamber algorithms is influenced by the accurate determination of the atrial rate [31]. The presence of atrial sensing errors and atrial blanking can result in either misclassification of ventricular arrhythmia as atrial arrhythmia or inappropriate rejection of ventricular arrhythmias.

The Atrioverter: Detection of Atrial Tachyarrhythmias

The philosophy of the atrioverter was different from a conventional ICD [32]. While ICDs had to recognize all ventricular tachyarrhythmias and were prompted to shock when in doubt, the atrioverter was allowed to wait until atrial fibrillation was diagnosed with absolute certainty. A specific device for atrial fibrillation, the Metrix atrioverter system (Incontrol Inc., Redmond, WA, USA), was developed with a two-step detection algorithm. The first algorithm discriminated between a sinus and a non-sinus rhythm, based upon the presence or absence of a 'quiet interval'. The second algorithm, the baseline crossing test (Figure 2.12), was performed on the electrogram obtained between the right atrial electrode and the coronary sinus lead (Figure 2.13).The result of both algorithms was a high sensitivity (100%) for the detection of non-sinus rhythm with a specificity of 96% for atrial fibrillation.

The standalone atrial defibrillator was safe, but required the implantation of a ventricular sensing lead. Mainly under influence of the industry, a dual-chamber device was developed that provided detection and treatment for atrial fibrillation, atrial tachycardia, and ventricular tachyarrhythmias [33].

In the majority of these devices, the detection of atrial arrhythmias is mainly based on rate (with sometimes overlapping zones for atrial tachycardia, flutter, and fibrillation). Evidently, this is not related to real physiology or pathology. For a more accurate classification of atrial tachyarrhythmias, another advanced atrial detection algorithm was developed [34]. This algorithm uses the maximum atrial rate, the standard deviation, and the dispersion of atrial rate to classify unstable and stable atrial arrhythmias.

Figure 2.12. Atrial fibrillation, as the number of baseline crossings is high.
Figure 2.13. Atrial fibrillation as detected by the Metrix. Wide band atrial and bipolar near-field right ventricular electrograms.

Morphology-based Algorithms

Some morphology-based algorithms existed in early devices (PDF, Intec and TPM, CPI) as mentioned earlier on. They were never widely used in this era. Medtronic based an algorithm on width of the intracardiac EGM and the slew rate. The ventricular EGM is then classified as narrow or wide by comparing the measured actual width to the programmed value for width threshold value (Figure 2.14). The optimal EGM source for measurements of the EGM width is a far-field configuration between can and coil. The efficacy of the EGM width criterion has been studied by several investigators (Table 2.2). The reported overall sensitivity for detection of ventricular tachycardia was 64.1% [35]. When the data was corrected for stable QRS-complexes in the 12-lead electrocardiogram, the sensitivity was higher on a per-episode basis than on a per-patient basis [35].

The EGM width criterion has several limitations. The criterion cannot be applied in patients with a pre-existing complete bundle branch, which results in underdetection of ventricular tachycardias. Rate-related changes of the electrogram width may result in false-positive detections of ventricular tachycardia. The EGM width is also affected by additional antiarrhythmic drug treatment (class Ic drugs and amiodarone).

Recently, arrhythmia discrimination in the ICD included more advanced morphology algorithms. These algorithms are based on more complex

Figure 2.14. Ventricular tachycardia with bipolar ventricular electrogram and marker channel. The QRS complexes are similar to the template which is shown in the lower part. Three out of eight complexes preceding the ATP episode are coded as 'Wide'. The criterion was only 'passive' in this case, as detection occurred because of the high rate.

Figure 2.14. Ventricular tachycardia with bipolar ventricular electrogram and marker channel. The QRS complexes are similar to the template which is shown in the lower part. Three out of eight complexes preceding the ATP episode are coded as 'Wide'. The criterion was only 'passive' in this case, as detection occurred because of the high rate.

comparisons of electrograms. The 'morphology discrimination' algorithm constructs a quantitative representation of each ventricular complex [36]. This representation is determined by the peak amplitudes, polarities, number of peaks, and the order of peaks. Each ventricular complex during tachycardia is compared to a stored template during the patients' baseline rhythm. The other morphology-based algorithms are the 'vector timing and correlation' algorithm and the 'wavelet transform' algorithm, which are both under clinical investigation [37, 38].

 Table 2.2. Efficacy of the EGM width criterion. Investigator Year Sensitivity % Specificity % Gillberg et al.