According to the figures released by World Health Organization (WHO), cardiovascular disease is the number one cause of death in the world. In France every year more than 50,000 people die suddenly due cardiac arrhythmias. Identification of high risk sudden death patients is still a challenge. To detect cardiac arrhythmias, currently Holter is generally used to record 1~4 leads electrocardiogram (ECG) signals during 24h to 72h. However the use of Holter is limited among the population due to its form factor (not user-friendly) and cost. An integrated single chip wearable Holter named SoC-Holter that enables to record 1 to 4 leads ECG is introduced. Deployment of wireless sensor network requires each SoC-Holter with less power consumption, low-cost charging system and less die area.To minimize energy consumption and system cost, Complementary Metal Oxide Semiconductor (CMOS) technology (0.35μm) is used to prototype the first implementation of SoC-Holter. Then a novel method based on Compressed Sensing (CS) technique is introduced for solving the problems of power consumption and storage capacity of SoC-Holter. The main principle underlying this framework is to sample analog signals at sub-Nyquist rate and to classify directly compressed measurement into normal and abnormal state. Minimizing the wire connected electrodes to the platform can make the carrier more comfortable because two electrodes are attached closely on the surface of the body. Recording difference ECG is analyzed using Vectorcardiogram (VCG) theory. Experimental results show that an integrated, low cost, and user-friendly SoC-Holter is feasible. SoC-Holter consumes less than 10mW while the device is operating. It takes advantage of estimating parameters directly from compressed measurements, thereby eliminating the reconstruction stage and reducing the computational complexity on the platform. In addition, the framework provides compressed ECG signals without loss of information, reducing significantly the power consumption for message sending and memory storage space. The effect of electrode placement is evaluated by estimating QRS complex in recorded ECG signals by two adjacent electrodes. The method is based on the QRS-VCG loop alignment algorithm that estimates Least Square (LS) between an observed VCG loop and a reference loop with respect to the transformations of rotation and time synchronization. The electrode location with less sensitive to interference is investigated.
