A Novel Adaptive Algorithm for Holter ECG Data Compression

Holter electrocardiogram (ECG) is the recording of the electrical activity of heart in order to assess its functionality during day-to-day activities. This involves a large amount of data to be recorded from multiple recording points of the body for a longer duration. The storage and transmission of this large volume of data is a great challenge. In this work, we propose a new adaptive algorithm for data compression that uses both Principal Component Analysis (PCA) and Wavelet Transform to exploit the inter-channel and inter-signal correlations of Multichannel ECG (MECG). First, a row wise transformation of the data is done using PCA and then Wavelet Transform is applied along the column. The proposed technique adopts an adaptive thresholding function to choose the error level in the reconstructed signal. The new coefficients thus obtained are quantised and Huffman encoded to obtain the compressed data. Using this algorithm, we were able to obtain a compression ratio of 16.83:1 with an APRD value of 6.59.