A Kalman filtering based adaptive threshold algorithm for QRS complex detection

This work presents an adaptive threshold algorithm in electrocardiogram signal feature extraction by introducing Kalman filtering theory. Low computational cost, low storage requirement and fast response feature are achieved by applying two sets of adaptive threshold systems in different conditions. Also, double-threshold peak detection design dramatically decreases the false detection conditions resulting from noise. As a proof of concept, the proposed algorithm is verified in Matlab and implemented on field programmable gate arrays (FPGA) using MIT/BIH database. The experimental results demonstrate proposed algorithm consumes low resource of FPGA and exhibits 99.30 % detection sensitivity and 99.31 % positive prediction in average, respectively. With the self-adjusting system, proposed algorithm can rapidly adapt different individuals in satisfied detection accuracy.