Rapid and coding-efficient SPIHT algorithm for wavelet-based ECG data compression

Abstract The set partitioning in hierarchical trees (SPIHT) algorithm is an efficient coding scheme widely used in wavelet-based electrocardiography (ECG) data compression systems. Traditional SPIHT schemes using three lists and a complex sorting process are unsuitable for wearable device design due to their time-consuming nature and low storage efficiency. In this paper, based on a bit-plane representation of quantized wavelet coefficients, a modified SPIHT algorithm is proposed for fast lossless coding. The bit plane is first represented by a tree data structure consisting of two types of primitive trees. The number of primitives is relative to the number of input sampled data. The coding process of the bit-plane data can be regarded as a synthesis that assembles the primitives sequentially based on various decision rules. The assembly rules are further simplified into logic-level decisions in terms of a flag scheme. Using the MIT-BIH arrhythmia database, the experimental results show that the proposed algorithm can achieve an up to 64.35% coding time reduction in comparison with the traditional SPIHT algorithm at a cost of a 0.28% bit-rate increase. The proposed algorithm is simple, regular, and modular.