Signal Adaptive Threshold for ECG Signal Compression Using False Discovery Rate Approach

This study proposes a noise-insensitive signal adaptive threshold for ECG compression that overcomes restrictions in earlier methods. The method employs a wavelet-domain adaptive threshold based on false discovery rate (FDR) measurement, which links hypothesis testing to thresholding. The FDR error control technique determines the false discovery threshold (FDT) based on the signal by computing and ranking the probability of each detail coefficient. Here, the Benjamini–Hochberg (BH) process is employed for implementation. The denoised ECG data are further compressed using Huffman coding and run length encoding (RLE). The method is suitable for thresholding and provides a noise-insensitive threshold. The reconstructed signal quality is outstanding, and the technique achieves a compression that is comparable to that of conventional codecs. The quality of the signal is evaluated by the mean structural similarity index (mSSIM), which is found to be almost one, suggesting a highly similar reconstructed ECG to the original ECG. Additionally, it is noted that the suggested method produces a lower PRD value, indicating improved reconstruction quality.