The CULT double-dictionary ECG compression technique achieved high compression ratios (5 to 98) with low reconstruction errors (RMSE 1.8% to 7%) and lower energy consumption compared to existing methods.
The proposed CULT algorithm provides an efficient, noise-immune ECG compression technique for IoT healthcare devices using a double-dictionary learning scheme.
The Internet-of-Things (IoT) healthcare system monitors a patients' condition and takes preventive measures in case of an emergency. The electrocardiogram (ECG) that measures the electrical activity of the heart is one of the important health indicators. Thanks to the wearable technology, nowadays, we can even measure the ECG using smart portable devices and send via a wireless channel. However, this wireless transmission has to minimize both energy and memory consumption. In this article, we propose CULT-an ECG compression technique using unsupervised dictionary learning. Our method achieves a high compression rate due to the essence of dictionary learning and is immune to the noise by integrating discrete cosine transformation. Moreover, it continuously expands the dictionary when the unseen pattern occurs and refines the dictionary when new input arrives, by imposing the double dictionary scheme. We show that our method has a better performance by comparing it with the other existing approaches.
Qian et al. (Tue,) conducted a other in Arrhythmia (ECG signals). CULT (double-dictionary based ECG compression technique) vs. TASOM, GSVQ, LTC, and pure DCT was evaluated on Root Mean Square Error (RMSE) and Compression Ratio (CR). The CULT double-dictionary ECG compression technique achieved high compression ratios (5 to 98) with low reconstruction errors (RMSE 1.8% to 7%) and lower energy consumption compared to existing methods.