Estimating residual starch from in vitro fermentations by Fourier transform mid-IR transmission spectroscopy

• Published in: Animal feed science and technology Vol. 152; no. 1; pp. 133 - 140
• Main Author: UDEN, P
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 2009; [New York, NY]: Elsevier Science; Elsevier
• Subjects: Agricultural Science; Biological and medical sciences; estimation; Feed and pet food industries; Food industries; Fourier transform infrared spectroscopy; Fourier transform mid-infrared spectroscopy; FTIR; Fundamental and applied biological sciences. Psychology; In vitro; in vitro digestion; in vitro studies; Jordbruksvetenskap; rumen fermentation; rumen fluids; ruminant nutrition; Solanum tuberosum; starch; Starch analysis; Triticum aestivum; Zea mays; SS; SD; CV; MPE; ISO; SD reg; FTIR; RMSPE; Starch analysis; In vitro; MSPE; Infrared spectrometry; Fourier transformation; Starch; Transmission; Carbohydrate; Fermentation; Polysaccharide
• ISSN: 0377-8401; 1873-2216; 1873-2216
• DOI: 10.1016/j.anifeedsci.2009.04.003

Fourier transform mid-IR transmission spectroscopy (FTIR) was evaluated as an alternative to enzymatic analysis of starch remaining after fermentation
in vitro. Different starch sources were incubated
in vitro for 0–24
h. Potato starch, raw or gelatinized, wheat starch and maize starch were used in the Calibration set. For the Test set, 6 wheat samples of different varieties were included. Observations from 198 and 98
in vitro tubes were included in the Calibration and Test sets, respectively. The analytical reference method consisted of an enzymatic procedure including a heat stable amylase, amyloglucosidase and determination of total glucose. Samples for FTIR analysis were all taken from the supernatant after the amylase hydrolysis step and titrated to pH 8.3. An FTIR instrument, normally used for milk analysis, was used to obtain sample spectra. Absorbance values from 153 wavelengths (6.3–10
μm) were used for calibrations against enzymatically determined starch expressed as glucose (g/L). Analysis of the Calibration set resulted in a multiple regression with five wavelengths with the highest slope parameter at 8.669 and the lowest (most negative) at 8.498
μm. Mean prediction error (MPE) was (0.058) and the determination coefficient (
R
2) was 0.994. The calibration equation also predicted the Test set well, as indicated by a high
R
2 (0.995). Mean prediction error was 0.125, but was mainly caused by bias and slope errors. The random error contributed to only 0.13 of total mean square prediction error which can be translated into a standard deviation of <30
g/kg air dry sample. Starch analysis by FTIR is precise and will simplify the analytical procedure and reduce the cost of analysis considerably.