The Kscore method relies on point-by-point spectral analysis of the T-wave magnitude at a frequency of 0.5 cycle per beat. Initially, each T-wave is measured at specific time form the QRS complex. Then, the fluctuations on the magnitude of each sample in T-wave vector is measured using the fast Fourier transform (FFT), generating a power spectrum for each point in the the T-wave. Finally, the power spectrum are aggregated and the final markers calculated using the two equation mention below:
Where, P corresponded to the accumulated amplitude of the power spectrum, usually at a f=0.5Hz. P_{N} and \sigma_{P_{N}} corresponds respectively to the mean and standard deviation of the power spectrum at the noise band, usually set at f=[0.43:0.49]Hz. The Kscore measures the statistical significance of the alternans in relation to the noise levels while KTWAM is a direct measure of the TWA magnitude. A visual representation of the algorithm is shown in the figure below.
ECG signal loading is supproted for pgysionet like formated files and requires the WFDB toolbox for Matlab: https://physionet.org/content/wfdb-matlab/0.10.0/