S. Atanassova

Near-infrared spectroscopy for biomonitoring: influence of somatic cell count on cow's milk composition analysis

The influence of milk with high somatic cell count on the accuracy of near-infrared spectroscopic determination of fat, protein and lactose content of non-homogenized milk was investigated. Transmittance spectra of 258 milk samples were obtained by the NIRSystem6500 spectrophotom eter in the 700–1100 nmregion. The best accuracy for fat, protein and lactose content determination was found for calibration equations, derived from samples with low somatic cell count (SCC). The standard error of prediction increased and the correlation coefficient decreased significantly, both when equations, derived fromsam ples with low SCC milk were used to predict the content of the examined components in samples with high SCC, and when equations, obtained for samples with high SCC were used to predict the content of the components in samples with low SCC. Therefore, milk samples with high SCC in a data set used for calibration or prediction highly influenced the accuracy of fat, protein and lactose determination. r 2001 Elsevier Science Ltd. All rights reserved.

Prediction of dry matter and protein degradability of forages by near infrared spectroscopy

Abstract The crude protein (CP) and dry matter (DM) degradability characteristics of 34 forage samples from various origin were determined in sacco using cows. NIR spectra of the samples were obtained by a NIRSystem 4250 spectrophotometer and the calibration equations were developed by partial least square (PLS) regression. Despite the limited number and heterogeneous nature of the samples, NIRS showed potential to predict all CP and DM degradability characteristics of the forages.

The possibility of near infrared spectroscopy for evaluation of microbial nitrogen content of in sacco feed residues and duodenal digesta of sheep

This study aimed to estimate by near infrared (NIR) spectroscopy the microbial nitrogen content (MN) of feed residues from in sacco degradability trails and duodenal digesta of sheep. NIR spectra from 50 samples of duodenal digesta, and from in sacco residues—110 samples of alfalfa hay and 38 samples of maize silage were obtained using an NIRSystems 4250 spectrophotometer. The microbial nitrogen (MN) content of part of the alfalfa hay in sacco residues (78 samples) was calculated from the percentage of 15N enrichment compared to enrichment in the original samples; for the rest of the alfalfa samples and samples of maize silage residues were determined by diaminopimelic acid (DAPA) as a bacterial marker, and MN of duodenal digesta samples by the purine N (RNA equivalent) content as a microbial marker. The calibration equations were developed by modified least squares as the calibration method. The microbial content of all kinds of samples was accurately calibrated and cross-validated. A standard error of cross validation (SECV) of 0.418 g microbial N kg−1 DM, a coefficient of determination for the cross validation of 0.925 and a ratio of standard deviation of population and the SECV of 3.88 were obtained for the alfalfa 15N labelled hay residues. For maize silage residues, the corresponding values were 0.832, 0.938 and 3.90, and for duodenal digesta samples the values were 1.05, 0.962 and 5.19, respectively. Prediction of MN as percentage of total N of the samples gave approximately the same level of accuracy. For example, the SECV was 2.35% units, cross-validation R2 was 0.953, SD/SECV was 4.60 for alfalfa 15N labelled hay residues. Despite the different origin of the analysed samples (feed residues and duodenal digesta), the NIR spectroscopy determination of MN content of all samples was based on spectral data at very similar wavelengths. The study indicated that NIR spectroscopy has the potential to predict microbial nitrogen content and to distinguish MN from total N content of in sacco feed residues and duodenal digesta.

Near infrared spectra and soft independent modelling of class analogy for discrimination of Chernozems, Luvisols, and Vertisols

The objective of the present study was to use near infrared diffuse reflectance spectroscopy and chemometrics to distinguish three major soil types in Bulgaria – Chernozems, Luvisols and Vertisols. The diffuse reflectance spectra of 78 air-dried and sieved soil samples collected from grasslands and arable agricultural lands and classified as Chernozems, Luvisols and Vertisols were obtained in the spectral range 700–2500 nm. Second-derivative transformation of soil and clay spectra was performed in order to find soil spectral differences based on the clay minerals content. Soft independent modelling of class analogy (SIMCA) was performed to classify samples according to their classes. Results show that the samples can be distinguished according to their classes as the SIMCA models correctly classified 100% of the samples in both calibration and validation sets. According to the discrimination power parameter of the SIMCA procedure, the most important wavelengths in the classification models of Chernozem, Luvisol and Vertisol soils were related to the content of clay minerals such as kaolinite and montmorillonite, smectite minerals, organic matter and carbonates.

Detection of bacterial contamination in milk using NIR spectroscopy and two classification methods - SIMCA and Neuro – Fuzzy classifier

Abstract Potential of near-infrared spectroscopy combined with multivariate classification methods for detection of bacteria in cow milk was investigated. Spectra of milk samples were obtained in a region from 600 to 1880nm. Presence of Staphylococcus aureus, Streptococcus agalactiae and other bacteria from genus Streptococcus , one of bacteria, causing mastitis in cows, was obtained in some of the milk samples by classical microbiological methods. One hundred milk samples negative for bacteria (class healthy) and one hundred milk samples with presence of bacteria (class contaminated) were used in the investigation. Two classification methods - Soft Independent Modeling of Class Analogy (SIMCA) and adaptive Neuro – Fuzzy Inference System (ANFIS) were implemented. SIMCA develops models for each class based on principal components (PC) that describe the variations of the spectral data. Once each class has its own model, new samples could be classified to one or another classes according to their spectra. The inputs to ANFIS were several principal components. ANFIS had only one output node, the class type. One-half of samples from each class were used as a training set for creation of SIMCA models or trained the ANFIS. The rest of the samples were used as test set for verification the obtained classifiers. SIMCA models, based on 7 PC correct classified 90% of samples from class contaminated and 88% of samples from class healthy for training data set. The results for testing the models with samples from test set were as follow: 90% of samples from class contaminated and 86% of samples from class healthy were correct classified. The average testing error for ANFIS classifier was 0.058% for class healthy was 0,032 % for class contaminated. Results of the presented experiments showed possibility to establish classifiers for identification of raw milk samples, infected with bacteria. Near infrared spectroscopy in combination with multivariate classification techniques offers an alternative approach to traditional methods with large potentials for a rapid and reliable application in microbiology, bio-diagnostics and food control.

Near-Infrared Spectral Informative Indicators for Meat and Dairy Products, Bacterial Contamination, and Freshness Evaluation

Abstract Near-infrared (NIR) spectroscopy opens a new area in biological sciences and engineering by exploring and describing biological systems through a nondestructive monitoring of their interaction with NIR light. Real application of NIR spectroscopy as accurate, fast, and noninvasive method for monitoring of growth of bacteria in meat and milk are presented. Modern spectral instruments and methods of data processing are applied to distinguish infected from noninfected meat and dairy products as well as to perform qualitative and quantitative determination of available bacteria, as an indicator of freshness or spoilage. Mastitis (intramammary infection) is a major problem for the global dairy industry. NIR spectroscopy describes informative patterns in application for somatic cell count determination, distinguishing of healthy and mastitis milk as well as in detection the presence and growth of specific bacteria, based on milk chemical composition changes and their influence on water spectral pattern.

Recognition of Fusarium diseased corn kernels using MIR spectroscopy and Neuro – Fuzzy classifier

Abstract In this paper we present the results obtained recently by recognizing corn kernels (healthy and Fusarium diseased) using spectral characteristics and Neuro-Fuzzy classifier. The proposed approach to corn kernel type diagnostic is based on three steps: spectral data acquisition, preliminarily processing and feature extraction, and classification. In the first step, the near infrared reflectance spectral characteristics are obtained in the range 400 – 2500 nm with step 2 nm. In the second step informative features are extracted using principal component analysis. In order to make a compact spectral data representation, the number of features is reduced. Adaptive Neuro – Fuzzy Inference System (ANFIS) is implemented in order to establish decision boundaries in the space of the selected features which separate samples belonging to two different classes (healthy and Fusarium diseased corn kernels). The inputs to ANFIS are several principal components. ANFIS has only one output node, the corn kernel type. The system is trained by data of train sample. We then evaluate the deviation of our estimate compared to the expert assessment. The procedure is applied to the 1800 corn kernels spectral data and the average accuracy is given. The average testing error for class healthy is 0,023 % and for class diseased is 0,329 %. As a result of the presented experiments the NIR spectroscopy is applicable as a nondestructive tool for Fusarium diseased corn kernels recognition and may become a powerful tool for monitoring the safety of our food supply. For practical implementation in the grain industry, a ANFIS classifier can be used effectively.