Development of PLS Model for Rapid Estimation of Protein Content of Rice Using Fourier Transform - Near Infrared Spectroscopy
Razvoj PLS modela za brzo određivanje sadržaja proteina u pirinču upotrebom furijeove transformacije – bliska infracrvena spektroskopija
View/ Open
Author
Pandiselvam, Ravi
Thirupathi, Venkatachalam
Mohan, Striramasarma
Uma, Doraiswamy
Keywords
NIR spectroscopyrice
protein
wave number
chemometrics
PLS model
NIR spektroskopija
pirinač
protein
broj talasa
hemometrija
Metadata
Show full item recordAbstract
This study attempted the feasibility to use near infrared spectroscopy as a
nondestructive analysis method to qualitative and quantitative assessment of rice quality
of Central Warehousing Corporation, India. A PLS model were developed using rice
standards of different concentrations in the near-infrared region (4.000–12.000 cm-1).
The developed models ware authenticated using test validation technique. FT-NIR
spectroscopy with chemometrics, using the PLS–first derivative plus vector
normalization method could predict the protein content of stored rice samples accurately
up to an correlation coefficient (R2) and residual predictive deviation (RPD) values were
0,98 and 7,21, respectively. The error values such as root mean square error of cross
validation (RMSECV) and root mean square error of estimation (RMSEE) were 0,28 and
0,25, respectively, with 11 factors in the prediction model. The developed model was
applied to predict protein content in rice samples within 15 seconds. The developed
procedure was further validated by recovery studies by comparing with micro Kjeldahl
method of protein determination. These results show that NIR spectroscopy could hold
up traditional techniques in studying qualitative assessment of rice. Ovim istraživanjem je primenjena mogućnost upotrebe bliske infracrvene
spektroskopije kao nedestruktivne analitičke metode za kvalitativnu i kvantitativnu
procenu kvaliteta pirinča. PLS model je razvijen upotrebom pirinča sa različitim
koncentracijama u bliskoj infracrvenoj oblasti (4.000–12.000 cm-1). Razvijeni modeli su
provereni tehnikom validacionog testa. FT-NIR spektroskopija sa hemometrijom,
korišćenjem metoda PLS-prvog derivate plus vektroske normalizacije, može predvideti
sadržaj proteina uskladištenog pirinča, sa koeficijentom korelacije (R2) i rezidualnom
prediktivnom devijacijom (RPD) od 0.98 i 7.21, redom. Mere greške, kao srednja
kvadratna greška unakrsne validacije (RMSECV) i srednja kvadratna greška procene
(RMSEE) bile su 0.28 i 0.25, redom. Razvijeni model je bio primenjen za procenu
sadržaja protein u uzorku pirinča u roku od 15 sekundi. Postupak je dalje ocenjivan
poređenjem sa mikro Kjeldahl metodom određivanja proteina. Ovi rezultati pokazuju da
NIR spektroskopija može da podrži tradicionalne tehnike kvalitativne analize pirinča.