Adapted-basis spectral transformations for "superimposed" peaks separation and mass spectrometric signal filtration

The paper considers the problem of using adapted-basis spectral transformation in order to improve the spectrum resolution and filter mass spectrometric signals. As reference signals necessary to construct the basis, various functions fitting the mass spectrometric peaks are used. The simplest function used to fit mass spectrometric signals is Gaussian. In addition, mass spectrometric peaks can be fit with flat-top functions and functions describing signals distorted by the detection system with an electrometric amplifier having a large time constant. The paper demonstrates that the direct and inverse spectral transformations in the basis "adapted" to the peak shape make it possible to reliably recover parameters of individual peaks contained in a mixed signal in case different peaks can be fit by similar functions different only in the amplitude and position on the independent variable axis. The recovery quality does not depend on the peak width and number of data in the sample. When filtering noise by spectral transformation, one can obtain well smoothed peaks whose positions can be precisely estimated in case the signal/noise ratio is no less than three.