Real-time filtering adaptive algorithms for non-stationary noise in electrocardiograms
The computer programs implement the adaptive algorithms for real-time ECG signal filtering and numerical simulation for evaluation of filter effectiveness. The adaptive algorithms with complex use Hampel identifiers and Z-parameter are an author’s development. To launch a program, enter the name of the program and the ECG model signal in the command line. For example: ahzmthp.exe clean.txt. The test signal parameters (its length) and parameters of the filtering algorithms are read from the text file named as “filters.txt”. The program requests the additive and multiplicative noise variance, the probability and the amplitude of the spikes, and the number of realizations for statistical averaging of the calculated filter performance indicators. For example, via the “space” key enter: 0.0001 0 0 0 200, then press “Enter”. To apply filtering to a test signal which is read from a text file, select the menu item "Load from file" by pressing the key "6". The filter results are put in the “RESULT” subfolder. The filter efficiency estimates are written to the "MSE.res" and "SNR.res" output text files. The input signal has an extension “.x” (no noise), “.xn” (with simulated noise), “.xns” (with noise and spikes). The signals from filter algorithms outputs have the extension “.yf”. Also, files with functions of identifiers used to adapt the algorithm parameters to the local signal behavior and to the changes in the noise level and with adaptable filter parameters, and other intermediate signals are put to the “RESULT” subfolder. Input and output signals are presented in text format in the form of two columns, where the first column is the numbering of the samples, the second one is the corresponding values of the discrete signal. The first number in the first line is the signal length. To apply filtering to signals of different lengths, the first number that indicates signal length in the second line in the "filters.txt" file has to be changed accordingly. Correction of the filter parameters (in the "filters.txt" ) for ECGs with another sampling rate is recommended. The program was compiled by Free Pascal.