Low-Power ECG-Based Processor for Predicting Ventricular Arrhythmia
Abstract : This paper presents the design of a fully integrated electrocardiogram (ECG) signal processor (ESP) for the prediction of ventricular arrhythmia using a unique set of ECG features and a naive Bayes classifier. Real-time and adaptive techniques for the detection and the delineation of the P-QRS-T waves were investigated to extract the fiducial points. Those techniques are robust to any variations in the ECG signal with high sensitivity and precision. Two databases of the heart signal recordings from the MIT PhysioNet and the American Heart Association were used as a validation set to evaluate the performance of the processor. Based on application-specified integrated circuit (ASIC) simulation results, the overall classification accuracy was found to be 86% on the out-of-sample validation data with 3-s window size. The architecture of the proposed ESP was implemented using 65-nm CMOS process. It occupied 0.112-mm2 area and consumed 2.78-µW power at an operating frequency of 10 kHz and from an operating voltage of 1 V. It is worth mentioning that the proposed ESP is the first ASIC implementation of an ECG-based processor that is used for the prediction of ventricular arrhythmia up to 3 h before the onset.
? The sequence of electrical activities are reflected into the ECG waveform, and therefore, temporal dependencies naturally exist in this waveform.
? LSTM solves this issue by allowing to forget according to the actual dependencies which exist in the problem.
? The first part, i.e., local data, is specific to the patient and is helpful in increasing the classification accuracy due to existing similarities among the heartbeats of every patient.
? This method cannot be applied to RNNs because of the existing temporal dependencies in the model, i.e., the feedback loops in which carry previous information through time.
? Ventricular arrhythmia is an abnormal ECG rhythm and is responsible for 75%–85% of sudden deaths in persons with heart problems unless treated within seconds.
? The main benefit of such approach is that the impact of a person’s motion and his daily activities is dramatically reduced.
? The main disadvantage of the system is that it uses fixed search window with predefined size to locate S and T fiducial points, which is not suitable for real-time scenarios.
? This computationally intensive algorithm is designed for a different problem namely classifying ECG signals into rhythms such as Sinus and Bigeminy.
• To extract the features automatically and increase the heartbeat classification accuracy, deep-learning based algorithms including deep convolutional neural networks and recurrent neural networks have recently been proposed.
• The proposed algorithm employs RNNs because the ECG waveform is naturally fit to be processed by this type of neural network.
• Our proposed algorithm is evaluated using the same ECG signals that were employed in the previous works that conform to this standard.
• In contrast, our proposed method is designed from ground up to be lightweight, and hence, meets timing requirements for continuous execution on wearable devices with limited processing capacity.
? These methods have exhibited advantages in the detection of ventricular arrhythmia, they have some shortcomings.
? It was intended to investigate the performance of the system without introducing the strong biasing effect of a classifier.
? The system is verified to operate for different clock frequencies, and we have reported the performance for operating frequency ranging from 10 kHz up to 4 MHz.
? The performance of the technique is highly accurate and satisfactory for advanced ECG feature extraction highlights the capability of the system to detect and delineate the ECG signal in different real-life scenarios.
|