AIR2-CT94-1416 Developing Analytical Methods from Fluorescence Spectra and Video Images with Relevance to Food and Non-Food Agro-Industrial Engineering Using New Computer Based Evaluation Techniques (AFFLUENCE)

Contacts
Summary Information



To find similar Items, click on a keyword below:
AIR Cluster V - Speciality Chemicals : Fibre : Fine Chemicals : Straw : Vegetable Oil/Fat

	Proposal No:	AIR2-CT94-1416
	Date Prepared:	October 1996, April 1998, May 1999
	Source:	Project Progress Report Third Year Report Final Report

---

Project Progress Report

INTRODUCTION

This project aims to establish appropriate hardware and software for fast, sensitive and reliable quality control of agro-industrial processes using fluorescence based spectroscopy and video image analysis. It links experts in various areas of processing with instrument companies. The project considers both food and non-food applications covering sugar, cereals and other seeds, lignocellulose (paper and fibres) and meat. Suitable equipment has been obtained, set up and calibrated in order to initiate spectroscopy and image-analysis calibration, together with experimental and operational protocols. The tasks can be divided into two groups. The first relates to general techniques. The second looks at applications. This report does not cover the work on meat, but includes studies on cereals and oil-seed where food and non-food applications may overlap.

OBJECTIVE

In agro-industrial production systems, where efficiency has so far been considered mostly in financial terms, there is presently a growing need for reliable, rapid and cheap screening methods for quality control and environmental protection in each production step. The fluorescence phenomenon is unique because of its specificity (e.g. in detecting NADH, Shiffs bases, lignin and mycotoxins) which can be exploited both by spectrofluorimetry and by video image analysis combing specific chemical information with structural/physical. Fluorescence is up to 1000 times more sensitive than other spectroscopic methods.

The emergence of new methods in this area depends on the development of appropriate hardware and software to solve a series of problems related to the physical/chemical aspects of the fluorescence phenomenon in relation to spectra and image recording, calibration and evaluation.

In this project, the development of hardware/software for non-destructive fluorescence techniques will be based on the one hand on application of advanced computer techniques to reflectance and transmission spectroscopy and on the other hand on innovative video imaging with the aim to establish on line industrial quality control methods (task A): These techniques are applied in three major industrial areas (task B): Considering meat, fibres (e.g. pulping) and cereals (e.g. oilseeds and wheat) with regard to oxidation (meat), lignin composition (fibre), tissue composition (meat, cereals) and filth and damage (seeds). In cooperation with the instrument industry spectrofluorimetric and video sensors will be designed on the basis of the research and their markets investigated. The opportunities explored are tested in cooperation with the production industry.

RESULTS

Spectrofluorimetry: Two identical spectrofluorimeters (Perkin Elmer LS 50B) with computers were acquired and set up in Copenhagen and Athens. Using sugar as a model it has been found possible to predict the chemical composition of inpurities such as ash, amino nitrogen and colour using fluorescence spectra in the near infrared area. Aspects such as lamp age, temperature, slit width, scanning velocity were standardised enabling complete spectra to be applied to various specific components using appropriate software from one state of measurement to another. Using sugar obtained from six separate factories, it was shown that each sample had a unique set of fluorescence characteristics. To resolve the spectra and establish the chemical basis, compounds were separated from samples of thick juice from the same factories using capillary electrophoresis. It was found the electrophoresis pattern of the pure components correlated with the factory classification and with specific wavelengths of the spectra. The experience gained from fluorescence measurements of sugars was then applied to wheat straw pulp for which it was found that emission maxima were dependant on several chromophores in the pulp. Bleaching by hydrogen peroxide increased fluorescence while photo-yellowing of pulps was accompanied by reduced emission intensity at excitation wavelengths below 400 nm. The fluorescent microstructure of samples was investigated using a DIPEX scanning fluorescence microscope with ccd camera. Other observations ranged from simple fluorescence microscopy to the use of a high sensitivity TIDAS diode-array apparatus used to obtain fluorescence and UV spectra from flow-cuvettes and from probes.

Fluorescence video image spectroscopy: In order to design a prototype system, various experimental setups were investigated using the a cooled Photometrics ccd camera. For low magnification imaging, an Oriel lamp with 2 computer-driven filter wheels was linked to the ccd camera and a computer. Software was developed for image discrimination. Mixtures of Cassava, maize and pea flours and the pure samples were tested in a model system. The system was able to identify the mixing levels with a very high efficiency. For higher magnification the ccd camera was installed on a Leica fluorescence microscope. Fluorescence microscopy was used to photographically document impurities in oilseed lots. It was confirmed that various contaminants including weed seeds, straw and the cereals, etc have typical autofluorescence characteristics which confirm that these may be detected by the ccd camera. Such investigations are now underway with the equipment participant optimising hardware and software.

Fibre quality assessment: Work has been initiated to correlated a wide range of paper pulp quality parameters with specific fluorescence spectra.

Oilseed quality assessment: Cooperation between an oilseed growers association, a French research institute and the hardware manufacturer has resulted in an image-based device for detection of impurities in oilseeds based on fluorescence.

Cereal grain processing industries: Attempts have been made to exploit the autofluorescence characteristics of the botanical components (pericarp, aleuron, endosperm) of wheat to monitor the separation in wheat mill flour streams using a Dipex scanning fluorescence microscope, with detection of particles in a ccd-camera by image analysis software. However, this has not been commercially successful. A wide range of wheat flours with known chemical composition such as ash, protein and colour were investigated using the Dipex microscope and compared to direct reflectance measurements in the Perkin Elmer LS50B spectrofluorimeter. Results were evaluated using Unscrambler software. Excellent correlations with ash indicative for aleuron were obtained by both methods but the Dipex method was too laborious. New software is being designed to speed up the method.

---

Third Year Report

Introduction
The project aims to promote contacts between industries and universities throughout the EU to promote the use of whole spectral and image fluorescence information as screening methods for a wide range of agro-industrial quality control applications. The project includes development of software for use in fibre applications for the pulp and paper industry, using multivariate analysis to utilise data from very complex fluorescence spectra. This requires knowledge of chemometrics and software development, skills that have been transferred through a series of systematic courses and workshops, especially to the Greeks. As a result French and Greek participants successfully combined fluorescence spectroscopy and fluorescence imaging techniques in fibre applications. The following achievements were attained during the third period covered by this report.

Progress in hardware development:
The project is investigating two different hardware applications. The first is spectrofluorometry, using commercial equipment that has been used intensively for fundamental studies that assist in software development as well as being used in a range of applications as described below. The second aspect of hardware involves the production of a prototype fluorescence video image spectroscope, which is functioning well. This has been used to gather data which that forms the basis for development of new software for discriminant analysis. An automatic scanning spectrofluorometric microscope has been tested for evaluating ash, colour and fibre in wheat flour mill streams with positive results. Using external funds the project has also acquired a CCD-imaging spectrograph with capacity for taking images as well as local spectra from images. This enables the homogeneity of samples to be investigated with a very high spectral resolution. This also led to development of new software.

As a result of the experience gained in spectrofluorometry collaboration with a Danish instrument company, Delta Optics has been established. This resulted in the construction of an "on-line" filter instrument suitable for use in sugar, milk and microbiological industries. A prototype for image capture has been built, tested and optimised.

Progress in software development.
Two important software developments have been realised. A special versions of n-way Partial Least Squares (PLS), Tucker and PARAFAC algorithms which are ideal for handling complex fluorescence spectra, has been developed. Using this it has, for the first time, been possible to deconvolute excitation and emission spectra directly with PARAFAC. This enables resolution of the underlying spectra of pure compounds such as dopa and tryptophan from complex fluorescence spectra of impurities, that could have applications in sugar production. In order to further validate the mathematical separation, the PARAFAC algorithm has been used to study separation of the same material on column chromatography, where it can be used to demonstrate purity of the peaks. In the area of fluorescence imaging, discriminant analysis techniques have been used to evaluate complex images from fibre material.

Progress in applied work.
Meat and fish quality assessment for SFK, a company that supplies the slaughtering industry with ultrasound scanning machines in order to evaluate carcass quantity characteristics such as fat and meat content. The company wishes to develop screening techniques which can also monitor quality characteristics, such as predicting the water-holding capacity of meat, tenderness and other physiological parameters of importance as far as eating quality is concerned. The application of fluorescence and W-VIS-NIR probes, as well as the CCD imaging spectrograph, have been investigated for this purpose. Positive indications have been obtained, where properly calibrated systems have been used, but the correlations are not yet good enough for use in practice. Because meat is a non-homogeneous product, calibration is difficult. Ideally calibration analysis (e.g. for water-binding) should be performed for the meat sample which is measured by the probe in order to get an optimal calibration model. In order to this precisely on large scale, a 25MHZ Nuclear Magnetic Resonance Instrument (NMR) has been employed. A method is being developed that, at the same time, can measure total water and fat in meat plus water binding by NMR. Positive results have been obtained.

Fibre quality has been assessed using spectrofluorometric methods as well as fluorescence imaging. The purpose is to obtain screening methods for quality assessment in the paper pulp and paper industry. To facilitate this fluorescence information has been compared with results from high resolution NMR, in order to demonstrate the nature of chemical structures that may be detected by fluorescence spectroscopy.

Quality in the oil seed processing industries has been investigated in collaboration with the oil seed industry. The studies have been completed. Collaboration with a soap and fragrances manufacturer and has developed well, with fluorescence spectroscopy (evaluated by chemometrics and calibrated to sensory tests) has been investigated for predicting and monitoring oxidation in soaps.

Quality assessment for the cereal grain processing industries is being investigated using fluorescence spectroscopy, which is faster than the DIPD scanning microscopy. However, the project is also investigating new ways to use the DIPD microscope, for example to detect germ particles in flour.