RAMAN APPLICATIONS IN FOOD ANALYSIS - Raman Spectroscopy for Soft Matter Applications - page 209

the range of 700- 1800 cm − 1 . A 180 ° back - scattering arrangement was
used, and no correction for the spectral response was applied. A total
of 32 scans was averaged for each sample, and the resolution was
32 cm − 1 . Each sample was measured in two repacked replicates, and the
average values were used in the data analysis. Figure 8.9a shows the
raw data for the 100 alginate powders where three samples with low,
high, and medium M/G ratios are highlighted. These same three samples
will be monitored throughout the section. As can be seen from the
fi gure, the measurements have a high scatter background, obscuring the
chemical information. This has to be handled before quantitative analy-
sis. Also shown in the fi gure are the nonsquared correlation coeffi cient
between signals for all samples and the M/G ratio reference values at
one wave number. This shows that no single variable in the spectra is
suitable for quantifi cation.
Multiplicative scatter (or signal) correction (MSC) is one of the most
widely used preprocessing techniques in spectroscopy. MSC in its basic
form was fi rst introduced by Martens and coworkers in 1983 [25] and
further elaborated on by Geladi et al. [26]. The concept behind MSC is
that by using a simple, linear preprocessing technique aimed at correct-
ing spectroscopic artifacts/imperfections, undesirable scatter effect due

12

(a)

2
0

M/G ratio 0.49
M/G ratio 1.29
M/G ratio 2.10
Correlation
coefficient R

7

(b)

1

0

-1

800 1000 1200 1400

Wave number (cm -1 )

Figure 8.9 (a) Raw Raman spectra of 100 alginate powders and correlation coeffi -
cients for individual wave numbers with mannuronic-to-guluronic acid (M/G) ratio
reference value; (b) multiplicative scatter (signal) correction (MSC) Raman spectra
of the same sample set.