A Compression Algorithm for ECG Data Using Variable-Length Classified Template Sets

In this paper, a variable-length classified template sets based ECG compression algorithm for ECG measurement system is proposed. First of all, apply R-peak detection to the ECG data and the data will be segmented according to the detected R-peak, then cluster the heartbeat vector quantities of different R-R intervals and extract the variable-length heartbeat template sets. Divide the ECG data into three parts including template, residual error and template index by template matching, and use Huffman coding to encode the residual error finally. In decoding algorithm, use the stored template sets, residual error, template index and Huffman coding dictionary to restructure complete ECG data. The performance of algorithm is evaluated by using MIT-BIH Arrhythmia Database. The average of compression ratio is 3.05:1, which is 94.3% higher than Huffman coding.