Understanding and Interpreting Dominant Frequency Analysis of AF Electrograms

Jason Ng, Ph.D.; Jeffrey J. Goldberger, M.D.


J Cardiovasc Electrophysiol. 2007;18(6):680-685. 

In This Article

Dominant Frequency Analysis

Dominant frequency analysis is a powerful tool for analysis of atrial rate in AF. However, when using dominant frequency analysis to estimate activation rate, it is important that the strengths and limitations of the technique are well understood so that the results can be properly interpreted. The main benefit of frequency domain analysis is that it can be easily applied to the complex electrograms of AF; the complexity of AF electrograms, such as varying amplitude and morphology, can make manual interval marking difficult. Some signals, such as the atrial fibrillatory waves in the surface ECG, do not even have distinguishable features to mark. Figure 3A shows an example of a bipolar electrogram recording where the activation deflections change in morphology and amplitude. This tracing contains 13 activation impulses with an average activation interval of 158 ms, corresponding to an activation rate of 6.3 Hz. The sixth and seventh deflections are smaller in amplitude and could easily be missed by an automatic marking algorithm. Assuming both deflections were missed, the average activation interval would instead be 190 ms, corresponding to an activation rate of 5.3 Hz. The power spectrum at the bottom of Figure 3D has a dominant frequency of 6.5 Hz. Frequency analysis works in this example because a single sinusoid of 6.5 Hz could still well approximate the signal despite the high variation of amplitude. The robustness of dominant frequency analysis on recordings with significant amplitude variability was validated with simulated signals in a previous study.[19]

Other common properties of AF electrograms, such as high variability of the activation intervals and complex fractionation, unfortunately present as much difficulty for dominant frequency analysis as they do in time domain analysis because the signals are not easily characterized by a sine wave that corresponds to the frequency of activation.[19] Figure 5A shows an example of a bipolar recording with highly variable activation intervals. The dominant frequency of this example is at 10.7 Hz. However, a second and third peak are seen at 5.3 and 6.7 Hz with a very similar power as the dominant peak. A small change in the signal could easily cause a significant change in dominant frequency from 10.7 Hz to 5.3 or 6.7 Hz; it is unclear which, if any, of these frequencies correspond to the atrial activation rate. Figure 5B illustrates a recording with split potentials. The underlying rate appears to be roughly 5 Hz. However, in this case dominant frequency analysis treats the split potentials as independent activations, resulting in a dominant frequency of 9.7 Hz (equivalent to an activation interval of 103 ms). Similar situations may occur with other complex fractionated activation patterns[19] or if far field ventricular depolarizations or noise are present.[20] Dominant frequency analysis is also sensitive to phase changes in the signal.[19]

Figure 5.

(A) An electrogram recording showing irregular activation intervals and changing morphologies. The power spectrum has a dominant frequency at 7.5 Hz, but has a peak of 10 Hz with comparable amplitude. (B) An electrogram recording with split potentials. Dominant frequency analysis treats the split potentials as independent activations, thus resulting in a dominant frequency of 9.7 Hz (equivalent to an unlikely activation interval of 103 ms) despite an underlying rate of roughly 5 Hz.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: