QRS subtraction for atrial electrograms: flat, linear and spline interpolations

Abstract

The main objective of this article is to implement and compare QRS subtraction techniques for intra-cardiac atrial electrograms based on using the surface ECG as a reference. A band-pass filter between 8 and 20 Hz followed by rectification, and then a low-pass filter at 6 Hz are used for QRS detection. QRS subtraction was performed using three different approaches: flat, linear and spline interpolations. QRS subtraction affects the power of the signals but it normally does not affect the dominant frequency. The average power of the atrial electrograms after QRS subtraction is significantly reduced for frequencies above 10 Hz.

An interactive platform to guide catheter ablation in human persistent atrial fibrillation using dominant frequency, organization and phase mapping

“…Ventricular far-field activity present in the AEGs might appear as misleading frequency components on the atrial frequency spectrum, affecting the accuracy of both the identification of the DF peak and phase mapping [18]. Therefore, a QRST subtraction technique was performed to remove the effect of ventricular far-fields for the 2048 virtual AEGs as previously described [19].…”

Analysis of QRS-T subtraction in unipolar atrial fibrillation electrograms

“…In contrast, the proposed approach aims to match the actual length and timings of onset and offset for all individual QRS-T segments. For Ahmad's QRS cancellation approach [9], it can be observed that: (i) QRS onset and offset are based on a fixed window, which certainly does not reflect the dynamic changes in the ventricular depolarization; (ii) there is no cancellation of the ventricular repolarisation component; (iii) there is no alignment between the QRS fiducial point in the ECG and the local VA detected in each AEG signal; (iv) during the subtraction, an interpolation method is applied (spline, flat or linear) to connect the onset and offset points, so any AA that overlaps with the QRS segment is also removed.…”

“…In Figure 3a, QRS-T segments are highlighted in the original electrogram. Figure 3b presents an AEG signal after QRS cancellation [9]. Figures 3c and 3d show the AEG signal after QRS-T subtraction respectively using the ABS and the proposed approach.…”

“…Examples of QRS and QRS-T segmentation using the three different methods are presented in Figure 2. The first approach (Figure 2a) was implemented by Ahmad et al [9] and the QRS onset (marked with squares) and offset (circles) are marked, respectively, as 41.67 ms (50 samples) prior the QRS fiducial point, and 50 ms (60 samples) after the QRS fiducial point. Figure 2b presents the classical ABS approach: The QRS-T pattern length has the same length as the shortest RR interval [7].…”

“…An alternative technique recently published defines the QRS onset at 41.67 ms prior the QRS fiducial point and the QRS offset at 50 ms after the QRS fiducial point and avoids the problem of identifying a QRS template by replacing the signal over the QRS period with an interpolation, with the authors favouring a flat line interpolator [9]. The three limitations of this approach are: (i) the length of the QRS to be removed is arbitrary and might not correspond to the actual onset or offset of some QRS complexes, resulting QRS residuals on the 'clean' signal; (ii) it does not attempt to preserve the AA during the QRS, as that part of the signal is simply replaced with a flat line, for instance, and (iii) it does not remove the ventricular repolarisation component.…”