WWW.WAS.ORG • WORLD AQUACULTURE • DECEMBER 2012 23 (CONTINUED ON PAGE 22) The challenge to global aquafeed manufacturers in today’s economy is to operate as efficiently as possible, without compromising quality, and keeping costs in check. Using classical wet chemistry methods, analysis of feed ingredients and feed samples can take several hours to complete, requiring high-cost reagents and chemicals. Feed mills lacking on-site analytical laboratories may have to send samples to third-party laboratories, which is timeconsuming and costly. During such periods, the lag can curb mill efficiency by causing materials to be held until tests are complete. The global aquafeed industry is urgently in need of a strategic action framework for quality control. Near infrared spectroscopy is a tool that can be used for process control, quality assessment, identification of raw materials and process by-products, and for quantitative chemical analysis of complex mixtures (Hymowitz et al. 1974, Rinne et al. 1975). Near Infrared Reflectance Spectroscopy (NIRS) Spectrometry is a spectroscopic technique used to assess or quantify the concentration or amount of a given substance. Infrared spectroscopy measures inter-atomic bond vibrations at different frequencies. In organic chemistry, analysis of infrared absorption spectra shows the type of bonds that are present in the sample. The infrared portion of the electromagnetic spectrum is divided into three regions — near-, mid- and far-infrared — named in relation to the visible spectrum. The near-infrared (NIR) spectral region extends from 780-2500 nm. In NIR spectroscopy, the sample is illuminated with NIR light and the spectrum of reflected light is recorded. Harmonic vibrations occur at unique infrared frequencies depending upon the quantity of absorber (analyte), type of absorbing molecules present within the sample and the sample thickness and color (Dreassia et al. 1998). The NIR method relies on correlations between a substance and the sample spectrum. The sample spectrum is automatically compared with standard spectra held in a central database and results are delivered almost instantly. Standard spectra are generated initially by assigning lab analysis values to numerous samples. When calibrated with accurate reference data, the NlRS prediction errors could be lower than the standard deviation of conventional reference data (Williams and Norris 1987). Qualitative and quantitative near infrared (NIR) spectroscopic methods typically require the application of multivariate calibration algorithms and statistical methods (chemometrics) to model NIR spectral response to chemical or physical properties of the samples used for calibration. Hence, Near Infrared Reflectance Spectroscopy and Aquafeeds S. Ferosekhan*1, D.L. Prabu2 and P. Antony Jesu Prabhu3 NIR instruments operate by statistically correlating NIR signals at several wavelengths with the characteristic properties of the sample to be analyzed (Jordon 1996a). Moreover, each biological sample has a unique NIR spectrum. So, if two samples have exactly the same spectrum, it can be assumed that they have the same physical or chemical composition. If spectra are different, then samples are different physically, chemically, or both (Ruiz 2001). Chemometrics The actual numerical value of a specific analyte in a sample — such as protein, starch or fat — is mediated by a calibration approach known as chemometrics. Jordon (1996b) describes it as “a discipline with one foot in chemistry and the other in mathematics.” Chemometrics applies statistical methods viz. multiple linear regression, partial least squares and principle component analysis to spectral data and correlates the spectra with physical or chemical properties or factors, which are directly determined, rather than the analyte concentration itself (Ruiz 2001). From a practical point of view, there are two chemometric concepts, Global H (GH) and Neighborhood H (NH). Chemometrics involves the transformation of the twodimensional spectral data to multidimensional space by any one of various loading-score mathematics. The loadings represent the independent patterns in the set of data and the scores represent the proportion of each pattern in the spectrum of different samples. This means, fundamentally, the spectrum of each sample is given a mathematical description and can be visualized graphically as a three-dimensional cube. Applications of NIR Spectroscopy in Aquafeed Manufacturing Traditional application of NlRS in the analysis of agro-food products has focused on the development of predictive equations relating spectral data to nutritional components, including crude protein, crude fat, and starch. As more data is generated, the analysis becomes more accurate. Near Infrared Spectroscopy can be used in many applications throughout the aquafeed industry. Incoming raw materials can be analyzed to verify quality. Raw materials can also be analyzed to ensure proper ration supplementation, control diet formulation, and optimize least-cost formulation. Precise quality results are used to make accurate calculations for feed Near Infrared Spectroscopy can be used to verify quality of incoming raw materials to ensure proper ration supplementation, control diet formulation, and optimize least-cost formulation.
RkJQdWJsaXNoZXIy MjExNDY=