A new procedure to reduce the background noise of multichannel chromatograp
hic data, named SPEAK, has been developed. It is based on the assumption of
gaussian elution profiles, but it can also deal with non-gaussian peaks.
The method has been tested with simulated data matrices, containing high no
ise levels, and real data matrices obtained by a fast-scanning fluorescence
detector coupled to a HPLC system. Results indicate that SPEAK is able to
lower the background noise by a factor of about 4 - 5, without any signific
ant losses of the relevant information.