B. Vandenbogaert et al., QUANTIFICATION OF OVERLAPPING CHROMATOGRAPHIC PEAKS USING A MATCHED-FILTER, Chemometrics and intelligent laboratory systems, 25(2), 1994, pp. 297-311
A novel approach to the quantification of overlapping chromatographic
peaks is introduced. The sum of the models for the individual overlapp
ing peaks is taken as the signal model in a matched filter. Thus, the
signal intensities being the objective of the quantification procedure
become parameters in an overall signal model. These and, if necessary
, other parameters are adapted by a modified simplex algorithm optimiz
ing the maximum in the output of the matched filter. A prediction of t
he results can be made on the basis of a noiseless response surface th
at can be calculated from the models. When shapes and positions of the
peaks are known and only their intensities need to be estimated, a qu
antitative theoretical error estimation is possible. The results thus
predicted are considered optimal and are used as a reference in the ev
aluation of the results of a range of experiments using simulated data
containing two overlapping Gaussians and first-order band-limited noi
se. The proposed procedure works well, the quality of the results usua
lly being on or just little below the theoretical optimum. Under condi
tions of high overlap or a low signal to noise ratio, the experimental
results no longer follow a normal distribution and their quality is l
ower.