Advanced technique for detection and removal of pacemakers artifacts from electrocardiographic traces

With the prevalence of cardiovascular diseases on the rise, the global pacemaker market is experiencing significant growth. The market size, valued at USD 5.87 billion in 2023, is projected to reach USD 10.77 billion by 2034, growing at a CAGR of 5.68\% during the forecast period. This surge in pacemaker usage reflects the increasing need for advanced cardiac rhythm management and underscores the importance of studies focused on enhancing ECG analysis in the presence of pacemakers. In response to this growing demand for pacemaker technologies, the project focuses on the development of advanced algorithms for the detection and removal of pacemaker pulses from ECG traces. These algorithms are essential for improving the accuracy of ECG analysis, especially in patients with pacemakers, where standard interpretation techniques may be confounded by the artificial signals generated by the device. Addressing the complex challenge of pacemaker pulse detection in ECG traces, the project developed two distinct algorithms. Addressing the complex challenge of pacemaker pulse detection in ECG traces, the project developed two distinct algorithms. The first algorithm, building upon established techniques, demonstrated efficacy in identifying PM pulses in clear ECG traces but encountered difficulties in the presence of noise. To overcome this, the second algorithm was crafted, harnessing the unique rising edge characteristic of PM pulses. This innovative approach proved successful in both noise-free and noisy conditions, significantly enhancing the ability to detect arrhythmias in scenarios where conventional arrhythmia detection software was ineffective. This advancement paved the way for the introduction of the "Flattening Method" and the "Interpolation Method" – two approaches aimed at isolating and removing PM pulses from ECG traces, thereby improving the clarity and reliability of cardiac rhythm analysis in clinical practice. The Flattened Method, a simpler technique, involved flattening the PM pulse to the baseline level, effectively neutralizing its impact on the ECG waveform. However, this method did not preserve the natural ECG morphology. The Interpolation Method, on the other hand, employed linear regression and spline interpolation. This approach initially replaced the PM pulse with a straight line connecting its highest and lowest points, followed by spline interpolation. This method successfully maintained the continuity and natural pattern of the ECG waveform. The validation of the algorithms across different frequencies revealed significant insights into their performance. Testing at higher frequencies, specifically 128 kHz and 64 kHz, yielded excellent results, with the algorithms achieving near-perfect detection rates. At these frequencies, sensitivity (Se) and positive predictive value (PPV) both reached 100\%, indicating exceptional accuracy in PM pulse detection and removal. In contrast, at lower frequencies such as 32 kHz, 16 kHz, 8 kHz, and 4 kHz, the algorithms' performance decreased, with both sensitivity (Se) and positive predictive value (PPV) values declining notably. For instance, at 32 kHz, Se remained high at 100\% while PPV slightly decreased to 97.50\%. However, at 16 kHz, Se slightly decreased to 99.97\% and PPV more substantially to 80.25\%. The decrease was more pronounced at 8 kHz, with Se dropping to 97.37\% and PPV to 18.29\%, and at 4 kHz, Se further decreased to 80.66\% with PPV falling to a mere 1.56\%. This drop in accuracy at lower frequencies is attributed to the increased distortion of PM pulses, highlighting the importance of high sampling rates in achieving reliable and precise ECG analysis. The choice of different frequencies for testing reflects a comprehensive approach to assess the algorithms' efficacy under varying conditions, ensuring their robustness and adaptability in real-world scenarios.