Algorithms based on interpolating techniques introduced the least error with 16-bit data, while least-squares data fit methods performed best on less accurate 8-bit data.
The optimal digital differentiation algorithm for computing the derivative of left ventricular pressure depends on the quantization noise level, with interpolating techniques better for 16-bit data and least-squares methods better for 8-bit data.
Five commonly used algorithms for digital differentiation are evaluated to determine how they perform in the presence of 8, 12, and 16 bit quantization noise. The algorithms are compared on the basis of rms error between a model derivative of the left ventricular pressure waveform and the approximate results of each algorithm. Algorithms based on interpolating techniques introduced the least amount of error when 16 bit data were used while algorithms based on least-squares data fit methods performed best on the less accurate 8 bit data. Some of the band-limiting characteristics of the algorithms are also discussed.
Marble et al. (Wed,) reported a other. Digital differentiation algorithms vs. Alternative algorithms was evaluated on rms error between a model derivative of the left ventricular pressure waveform and the approximate results of each algorithm. Algorithms based on interpolating techniques introduced the least error with 16-bit data, while least-squares data fit methods performed best on less accurate 8-bit data.