Baseline variations

Visual inspection of key naturally occurring peaks from a gas chromatograph trace in relation to the usual baseline variation, reflecting the sensitivity or the... 

The effects of pixel-to-pixel baseline variation can be greatly reduced by taking the first or second derivatives of the spectra as shown in the section above however,... 

Phase 2 - data preprocessing. There are many ways to process spectral data prior to multivariate image reconstruction and there is no ideal method that can be generally applied to all types of tissue. It is usual practice to correct the baseline to account for nonspecific matrix absorptions and scattering induced by the physical or bulk properties of the dehydrated tissue. One possible procedure is to fit a polynomial function to a preselected set of minima points and zero the baseline to these minima points. However, this type of fit can introduce artifacts because baseline variation can be so extreme that one set of baseline points may not account for all types of baseline variation. A more acceptable way to correct spectral baselines is to use the derivatives of the spectra. This can only be achieved if the S/N of the individual spectra is high and if an appropriate smoothing factor is introduced to reduce noise in the derivatized spectra. Derivatives serve two purposes they minimize broad... 

A second factor that contributes to the baseline variation is the difference in the background signal (absorption fluorescence) between the two solvents. This effect causes the difference in the baseline level between the left and the centre in figure 6.6b. A more extensive discussion on baseline variations in programmed solvent LC can be found in ref. [607]. 

If a is some (standard deviation type) estimate of the baseline variation inherent in an industrial process (obtained, e.g., from a calculation such as (5-10) or from data taken from the process after eliminating all physical sources of assignable variation), it essentially specifies what is possible in terms of consistency of process output. There are, however, several common ways of using such an estimate to produce related measures of process capability. 

Fig. 3 Signal and noise measurements illustrated. The signal is measured from the apex of the peak to the middle of a straight-line peak base. The noise is measured as the distance between straight lines constructed from the tops and bottoms of the baseline variation. The noise should be measured in a clean portion of the chromatogram and should include a fair representation of the inherent baseline variation.

HIXKIN was tested for ability to reproducibly fit simulated and real data. For simulated data with added random noise, HIXKIN achieved good fits for first- and second-order cases (a and fi, respectively) as long as the added noise was less than 61 of the optical densities at the beginning of the decay (Table HID). Although HIXKIN achieved satisfactory fits of a, and fi for an artificial data set with several standard deviations of added baseline variation (Fig. 6 in 8), a, the more interesting term, was more heavily affected by baseline shifts than was 8, the mathematically dominant term at short times after the flash. 

Once the estabhshed requirements baseline variations and conflicts are satisfactorily resolved, the requirements baseline is considered valid. This validated requiremraits baseline is thrai used as input to functional analysis (6.3) and documented in the integrated database. 

As a rule of thumb, peak area calculations are sensitive to baseline variations, and therefore many trace analyses are quantified over the peak... 

Heiman, A. and Licht, S., Fundamental baseline variations in aqueous near-infrared analysis. Anal. Chim. Acta, 394, 135-147, 1999. 

Derivative Methods Derivative methods have long been used in NIR spectroscopy as pretreatment methods for resolution enhancement as well as baseline correction (24, 25). A derivative spectrum is an expression of derivative values, d A/dX" (n = 1, 2, ) of a spectrum A(X) as a function of A,. The second derivative, d AfdX, is most often used. The superimposed peaks in an original spectrum turn out as clearly separated downward peaks in a second derivative spectrum. Another important property of the second derivative method is the removal of the additive and multiplicative baseline variations in an original spectrum. On the other hand, a drawback of the derivative methods is that the SN ratio deteriorates every time a spectrum is differentiated. 

Figure 3.3 displays the second derivative obtained with the Savitzky-Golay method of the spectra shown in Figure 3.2A (19). It is noted that the calculation of the second derivative makes a number of bands clearly detectable. Moreover, it can be seen from the second derivative spectra that the second derivative is powerful in removing additive and multiplicative baseline variations of the spectra.

The equations based on spectral data transformed as second derivative or by using baseline correction showed lower accuracy, suggesting that the baseline variations, corrected by the second-derivative transformation or baseline correction, contained information that is significant for SCC determination. 

To minimize the effects of baseline variations due to Mie scattering, the ANN analysis was performed on second derivative spectra derived from the raw spectra processed using a Savitzky-Golay algorithm (nine points). 

---