Atrial fibrillation (AF) is currently a common clinical arrhythmia with a high morbidity and mortality but the electrophysiological and pathological mechanisms of AF are still not well understood [1–5]. The analysis of the atrial electrical activity plays an important role to reveal the underlying electrophysiological mechanisms responsible for AF’s initiation, maintenance and perpetuation. In recent studies, sites with high dominant frequency (DF) are regarded as drivers of AF [6–8] and therefore become the targets for AF ablation [7, 9, 10]. Hence, in ECG mapping system dominant atrial frequency is a key parameter for the analysis of AF [5, 11, 12].

For atrial activation, DF is the frequency component which marks the atrail dominant rhythm. In the frequency domain it is the maximum fundamental peak [8, 9, 13]. For example, if the myocardial tissue excited 30 times in 10 s, the DF should be 3 Hz, and the amplitude of 3 Hz should be the largest peak in the frequency domain. Botteron’s approach is the most widely employed preprocessing approach to estimate the dominant atrial rate. The preprocessing steps are: (1) band-pass filtering at 40–250 Hz, (2) absolute value, and (3) low-pass filtering at 20 Hz [8, 9, 14–16]. After these three steps, DF is the maximum peak’s corresponding frequency in the power spectrum wave. According to ISI Web of Knowledge, the number of research with title containing “dominant frequency” is about 140, and up to date the works of Botteron [14, 15] have been cited 225 times [17, 18]. When Botteron first proposed this method in 1995 and 1996 [14, 15], his approach was not for seeking the DF, but for the correlation calculation. However, it is now regarded as a classic preprocessing method [8, 9, 13, 16, 17].

In 2014 Castells et al. explained the theoretical basis of the preprocessing process [17], but they didn’t introduce what should be noted on real data analysis. Furthermore, in recent years, some researchers do DF analysis not in accordance with the above three steps [12, 19, 20]. In this paper the role and the necessity of each step in Botteron’s approach are addressed by using real data recorded from the 128-channel unipolar epicardial mapping system. For real time DF analysis, we also put forward two proposals to save the calculation time: reducing the sampling frequency and ignoring the ventricular artifacts.

In this study, we devised animal experiments to collect real-world data using living canine subjects with cholinergic AF model. The canine heart was exposed after thoracotomy and suspension of the pericardium. 8 flexible patches containing a total of 128 unipolar electrodes were carefully attached to right atrium, left atrium and the root of pulmonary veins. AF was induced by rapid pacing (with a frequency of 20 Hz) at the atrial appendages of the canine heart with an intravenous injection of acetylcholine (see details of experimental procedures and mapping sites in [21] and [22]). Then real data were recorded at a sampling frequency of 2 kHz from the 128-channel unipolar epicardial mapping system developed by the Electrophysiology Laboratory in Fudan University [21, 22]. The hardware of our mapping system included amplifying and filtering, and the hardware filters were a notch filter at 50 Hz and a band pass filter (BPF) at 3–500 Hz. The software which contain signal processing and analyzing were implemented with MATLAB (Mathworks Inc.). 8 dogs (weight 13.4 ± 2.9 kg) was used, all these data formed a signal database in the Electrophysiology Laboratory of Fudan University, and all the AF data used in this study were randomly selected from this database.

Dominant frequency and Botteron’s approach

Figure 1 shows a unipolar epicardial AF signal and its spectrum. DF is an important frequency domain index, which often detected as the frequency with the maximum power in the power spectrum. Preprocessing may be needed for a more accurate DF result. Botteron’s approach is just the classic preprocessing method, its preprocessing steps are: (1) band-pass filtering at 40–250 Hz, (2) absolute value, and (3) low-pass filtering at 20 Hz.

Figure 2 shows each step of Botteron’s approach and the signal’s spectrum after each step. The spectrum is based on fast Fourier transform (FFT). A Hamming window is used before FFT.

Each of the above three steps has its specific role. But considering the reason that this preprocessing algorithm was not for DF analysis, but for the correlation calculation when Botteron first proposed it in 1995, these three steps could be adjusted if used in seeking for DF, especially during AF.

Taking absolute value

Taking absolute value in Botteron’s approach is also known as the rectification to restore the low frequency components after band-pass filtering [17]. During this process, the DF which reflects the main rhythm of signal will appear and be highlighted. After taking absolute value, the amplitude of the direct current (DC) component is the largest and the maximum peak will appear at 0 Hz. Then, DF is shown as the second maximum peak. For the theoretical explanation of the effect by taking absolute, you could also see the article [17], which is to find out the changes in the spectrum area by the analysis of phase change. Here we will illustrate the role of absolute value from another perspective.

Due to myocardial cells’ fast depolarization, the epicardial ECG signal will have a steeper slope when the tissue is exciting. The shape of a single ECG after taking the absolute value is similar to that of waveform f(t) in graph (a) or (b) of Fig. 3, some may also be similar to the waveform shape in graph (f). From a theoretical point of view, one episode of epicardial ECG signal can also be obtained (mainly superposition) by the linear combination of waveforms in (a), (b), (c), (d) and (f). The spectrum of these five waveforms all decreases their amplitude when frequency increasing from 0 Hz as the red box in Fig. 3 shows. In case the waveform occurs periodically, the spectrum becomes discrete spectrum, as shown in Fig. 3f. The decreasing envelope shape, combined with discrete spectral line appears at every other f ₀ (f ₀ = 1/T). We can clearly see that, except 0 Hz, the first line has the maximum peak and is corresponding to the frequency of any periodical signal. This is the reason why the DF could reflect signal’s rhythm (period or frequency) and accords with the largest peak in the spectrum.

The role of taking absolute value is very important. When a signal has both positive part and negative part (before taking absolute value), as graph (e) shows in Fig. 3 [e.g. the ECG waveform in Fig. 6 is close to (c) plus (e) in Fig. 3], its spectrum would not show a single reduction trend from 0 Hz. Thus when the first spectral line falls into the decreasing area, it is still the largest peak, which can accurately correspond to the DF. Otherwise, when the first discrete line falls into the area before the maximum peak (the blue box in Fig. 3e), the maximum peak line may no longer corresponding to f ₀.

The preprocessing stage first used by Botteron and Smith is now considered the classical method for electrogram processing, especially for fundamental frequency detection. This paper aims to elaborate on the specific roles of those three steps for the analysis of dominant atrial rate.

In the first step the pass band range of BPF should be carefully chosen according to the signal energy distribution. We found that the best pass band for our data was 20–100 Hz, and 20–200 Hz could also get a good result, but if for other signal, this might change. Ciaccio et al. has found the optimal pass band for complex fractionated electrograms was 20–250 Hz [30]. Although it was done with bipolar signals and clinical data, its pass band was similar to our 20–200 Hz. Subsequently, the second step of taking absolute value is important in seeking DF, and we have analyzed its role from another point of view different from the paper [17]. In some cases directly doing FFT without taking absolute value and band-pass filtering can also get the correct DF result. This paper compares the two methods’ DF result on 128-channel epicardial mapping signals and finds that nearly half of the signals get the same DF result by the two methods. Nevertheless we highly recommend the nonlinear time-domain process of rectification. This is the key step to emphasize the fundamental frequency (Fig. 6). Finally, from the above results we conclude that the third step of 20 Hz LPF is completely unnecessary for the detection of dominant atrial rate.

The more regular the signal is, the more reliable the DF result will be. But for some AF signal with irregular and continuous electrical activity the DF results are not credible [9, 17]. In the experiment corresponding to Fig. 8, we find that on the whole the preprocessing method is a suitable method for AF mapping signals.

Furthermore, the sampling frequency does not affect the DF result. In real-time analysis, down-sampling could be used when doing the DF detection. It’s worth noting that the canine’s heart rate is higher than human being’s and we find that the activation rate of some signals is over 10 Hz. When the sampling frequency is decreased to 20 Hz, the largest frequency of the after-down-sampling signal is 10 Hz, it can’t get the correct DF value. This is why the accuracy rate of 50 Hz in Fig. 10 is 74.5 % but that of 20 Hz is only 49 %.

During AF ventricular artifact signals are flooded in atrial signals. If the ventricular artifacts are small, it may be unnecessary to remove ventricular artifacts when evaluating the dominant atrial rate. But with regard to body surface mapping, the impact of the ventricular signal may be taken into account because the amplitude of the ventricular signal is much larger than that of the atrial signal.

In fact, when seeking the DF value in signals with some types of artifact signals, it can also get the right result by Botteron’s approach even though the artifact is not removed. In 2013 Haissaguerre et al. used the DF of body surface mapping signal to identify localized sources of AF [31]. In their study, in order to keep the atrial component, the QRS wave is not removed even though it is much larger than the atrial wave. In this case, Botteron’s approach has the role of reducing the influence of ventricular artifact when evaluate the atrial DF. Please keep these points in mind, you will find ignoring the ventricular artifact is a feasible way for the detection of the dominant atrial rate. It is worth mentioning that the ventricular component becomes evident in unipolar electrograms with respect to bipolar recordings. However, unipolar electrograms may contain more useful information of atrial activities.

Compared to clinical recordings, canine model may have limited relevance to clinical recordings. Our model does not have changes in the substrate such as fibrosis which can lead to additional wavefronts, conduction block or discontinuities. These would add detail to the electrograms making the detection of DF more difficult.

The suitability of the Botteron’s approach employed in practical applications is discussed. It can be inferred from the above that the pass band of BPF can be adjusted depending on the energy density distribution and the rectification of absolute value is important. Nevertheless the 20 Hz LPF is totally unnecessary. In addition, during AF, based on the consideration of real-time analysis in clinical practice, we can use the down-sampling method and ignore the ventricular artifacts.