High-resolution detection of sustained ventricular and supraventricular tachycardia through FPGA-based fuzzy processing of ECG signal

Abstract

The paper presents a field-programmable gate array (FPGA)-based fast processing system with 12-channel high-resolution (24 bits) front-end for ECG signal processing. The implemented high-resolution data conversion makes the system suitable for recording of late potentials of the QRS complex in patients prone to sustained ventricular tachycardia. The system accepts ECG signals through 12 channels and then filtered to minimize baseline wander and power-line interference. The filter outputs are connected to 12 delta-sigma ADCs. The whole ADCs work synchronously at 8 kHz sampling frequency, and their output data are transferred to an FPGA that computes online on the digitized sample values in real time and ascertains whether the patient under study suffers from ventricular tachycardia or not. In order to ascertain the QRS complex accurately in the noisy ECG signal, fuzzy entropy of the sample values has been computed and provided as an input to inverse multiquadratic radial basis function neural network. Using the standard CSE ECG database, the algorithm performed highly effectively. The performance of the algorithm in respect of QRS detection with sensitivity of 99.83 % and accuracy of 99.7 % is achieved when tested using single-channel ECG with entropy criteria. The performance of the QRS detection system has been compared and found to be better than most of the QRS detection systems available in the literature. Using the system, 200 patients have been diagnosed with an accuracy of 99 %.

Methodological identification of anomalies episodes in ECG streams: a systematic mapping study

“…With the aim of this solutions, an increasing number of studies have found that relates the parametric importance of different ECG features [40]. According to literature, these parametric factors rely on external or internal variables, which can be referred to as feature dependencies [41,43,45]. Various state-of-the-art classification techniques for different T-wave patterns are sourced from the existing literature [58,59].…”

“…In literature, a number of studies found that build a reasonable accuracy level in terms of classification of different T wave episodes, but due to these methods, the difference between flattened T wave and inversion T wave is still unclear [30][31][32][33][34]. Accurate and robust classification of different T wave episodes relies entirely on the identification of the values of T wave parameters especially in terms of T-onset and T-offset [35][36][37][38][39][40][41]. Such findings are further helpful for highlighting the intensity of MI in terms of ST-segment elevation and ST-segment depression [42][43][44][45][46].…”

Review: Recent Directions in ECG-FPGA Researches

“…The suggested diagnostic system based on fuzzy neural is realized on Altera EP1C6Q240C8 FPGA platform to take advantage of FPGA low cost and flexibility to map changes in the proposed algorithm. According to the results obtained by [33], the applied fuzzy neural FPGA based realization was fast with very high detection accuracy comparing to other QRS methods detection.…”

“…In [33], present FPGA realization aims to detect QRS complex and utilized it for ventricular and supraventricular tachycardia diagnosis in real-time manner. The research applies the fuzzy entropy measure of high resolution ECG sample values to neural network.…”