Director Véronique Bellon-Maurel

Irstea

Could near infrared spectroscopy be useful to digital agriculture?

Tuesday June 13, 13.30 – 14.00

Digital agriculture deals with the use of Information and Communication Technologies in agriculture, including data acquisition, data management and data processing, including new approaches such as big data processing when data becomes massive (e.g. with satellite images). The aim of this chain is to turn data into pieces of information useful for the farmer or the farmer advisor, in order to make decisions easier and more relevant. NIR sensors (either spectroscopic or imaging ones) are good candidates to be the data acquisition devices at the start of this data-to-information chain. They have proven to be useful for characterizing agricultural product quality and soils, either manually (e.g. for fruit or animal) or embedded in farm machinery (e.g. wheat and fodder harvesting machines), but to detect stresses such as nutrition deficits or water stress at larger scale by use of airborne devices (airplanes, UAV) and satellite images. Several challenges still remain to tackle with by research involvements, to increase the role of NIR sensors for digital agriculture. They are related to difficult sensing issues, such as detecting disease before symptoms become visible, contributing to detect attacks by insects, improve water stress analysis, detect animal disease, analyzing soil quality with the aim of providing advices about sowing density and fertilization, etc and with the aim of the “4 per mille” carbon issue. They are also linked to the constraints put on the technological and economic constraints applied to precision agriculture sensors, ie being performing, robust and low-cost. New measurement strategies involving NIR sensors are therefore necessary, to enlarge the “information density” that can be brought by NIR sensors and imaging systems: special geometry of the optical probe, fusion of data from several NIR sensors or imagers (including series analysis), combination of NIR data and other sources of data, applying big data processing on accumulated NIR data (e.g. hundreds of thousands of spectra collected in the national grain analysis laboratories) and associated data (variety, soil type etc) to improve calibration models. These various opportunities for NIR sensors to spread even further in precision agriculture will be presented.

Biography: Véronique Bellon-Maurel, is presently director of the environmental technology department at Irstea, after having been for 10 years the director of ITAP joint research unit, where she developed her research, in close relationship with Jean-Michel Roger. With a basic background in agronomy, she got specialized in near infrared (NIR) sensors applied to biological products and precision agriculture, with a special involvement in probe optical design. She tackled both with academic issues (how to improve the signal / noise ratio by a better light/matter interaction control) and industrial concern (she is co-inventor of a NIR portable device to assess grape maturity). During a sabbatical stay at University of Sydney, in 2009, in Pr Alex Mac Bratney’s lab, she started working on NIR applied to soil, a subject that then rooted in Irstea, with the development of specific probes e.g. POLIS to get a better S/N ratio on such difficult media. She got deeply involved in ICNIRS community as being the founder and first president of HelioSPIR, the francophone society for NIR spectroscopy, the chair of NIR2013 in La Grande Motte, France, and member of the ICNIRS chairman’s advisory committee. In 2008, she was laureate of the Tomas Hirschfeld Award. In 2017, she launched #DigitAg, a multidisciplinary Convergence Laboratory dedicated to Digital Agriculture which promotes, inter alia, sensor development for precision agriculture including, of course, NIR sensor development.

Associate Professor Maria Angela Franceschini

Harvard Medical School Associate Physicist

Clinical Neuro-Monitoring with Diffuse Correlation Spectroscopy

Thursday June 15, 13.30 – 14.00

With the foundation of our seminal near-infrared spectroscopy (NIRS) and diffuse correlation spectroscopy (DCS) work on neonates, we are developing novel devices and approaches to better quantify cerebral blood flow and oxygen metabolism in the clinical setting both in children and adults. In particular, in the last few years we have made substantial progress on DCS technology to improve depth sensitivity, speed and accuracy of the recovered blood flow index (BFi). Validation studies, clinical applications and path to commercialization will also be discussed.

Biography: Dr. Franceschini is an Associate Professor at Harvard Medical School with specific training and expertise in the development of non-invasive optical techniques and applications in neuroscience, neurology, and brain health. She has been a member of the Optics Division of the Martinos Center at Massachusetts General Hospital since 2000. As a pioneer in the field of near-infrared spectroscopy (NIRS), she has made substantial contributions to the development of NIRS instruments and to the modeling and testing of diffusion theory to describe light propagation in turbid media. She has successfully applied the technology to a large range of functional neuroimaging and clinical neuromonitoring applications.

Dr. Steve Holroyd

Fonterra

The Use of NIRS in the Dairy Industry: New Trends and Applications

Monday June 12, 13.30 – 14.00

Keywords: NIR, dairy, compositional analysis, quality, authenticity, adulteration, food safety, contamination

Milk and milk products are a widely produced and traded source of nutrition globally. Milk itself contains a variety of components. Combined, these afford milk a healthy, nutritious profile that is key to its value as a globally recognised food. Of importance to all dairy processors is ensuring that this goodness is maintained throughout the value chain and is readily available to consumers. Near infrared spectroscopy (NIRS) is an important element in the quality systems that ensure that milk and milk products are a safe source of human nutrition.

NIRS has been used routinely to quantify the composition of milk and milk products for over 30 years. Initial applications were on powdered products and cheese immediately post processing but successive developments in hardware, software and chemometrics have widened the scope of application to include the entire range of dairy products throughout the production chain. Concurrently, approaches such as using instrument networking and in- and on-line analysers have further optimised the application of NIRS.

The last decade has seen a substantial growth in areas such as the application of NIRS to detailed compositional parameters that were previously determined by specialised assays. Proof of origin and authenticity testing as well as more qualitative applications, such as health benefits, have also been widely explored. Also, over this period, there have been a number of documented incidences of both deliberate adulteration and accidental contamination of foodstuffs. This has highlighted a clear need to detect these contaminations by fast, accurate and non-destructive methodologies. This has stimulated researchers to explore the possibility of applying NIRS for this purpose. NIRS has the potential for detecting the contamination of milk and milk powder both by economically motivated adulterants such as melamine and by contaminants that may have been introduced unintentionally.  However, applications such as these are testing the limits of sensitivity of NIRS and considerable care must be taken to ensure that the system set-up is robust and capable of providing effective quality assurance over extended periods. There are a range of factors that must be taken into account to ensure that this occurs.

Biography: Dr Steve Holroyd is a Research Programme Manager within Fonterra’s External Science and Technology team. His current responsibilities include leadership of research on analytical science and food processing. He has 25 years’ experience in the implementation of infrared analytical spectroscopy in oil & gas, pharmaceutical and food applications. On behalf of New Zealand he is involved with a variety of standards development organisations globally including IDF, ISO and AOAC. Steve has a PhD in chemistry from University of Auckland (NZ) and post-doctoral experience at Cambridge University (UK). Prior to joining Fonterra he worked for Texaco Ltd as an analytical chemist in the UK.

Prof. Dr. Christian Huck

Universität Innsbruck

Theoretical and technical advancements of NIRS and its operational impact in industry

Thursday June 15, 08.30 – 09.00

In the industrial environment, NIRS on one hand enjoys increasing popularity due to its defined advantages over classical analytical techniques (e.g., chromatography) including fast, non-invasive and simultaneous determination of several physical and chemical parameters. On the other hand, hardly any other analytical discipline is making such fast and fundamental technical and theoretical advancements. Miniaturization of spectrometers down to light weight portable devices is achieved via micro-electro mechanical systems (MEMS) or linear variable filter (LVF) instruments, operating with restricted performance in a defined wavenumber range. Fast NIR imaging devices are becoming popular checking conformity and spatial distribution of ingredients. NIRS is also benefiting from theoretical advances including design of experiment (DOE) for systematic optimisation purposes, special algorithms for calibration transfer between different types of spectrometers, and last but not least, quantum chemical approaches are becoming essential for distinct band assignments especially in case of complicated combination and overtone vibrations. Care must be taken about how these advancements can be successfully applied in the industrial environment.

Biography: Christian Huck currently holds the position as a full professor at Leopold-Franzens University, Innsbruck, Austria and is the vice head of the Institute of Analytical Chemistry and Radiochemistry. He is chief editor of NIR News and editor of Journal of Near Infrared Spectroscopy (JNIRS), has published more than 200 publications in the field (h=32) , holds 14 patents and is the receiver of several international awards. Since 2015 Huck also works as a guest professor in Japan at Kwansei Gakuin University (Sanda, Hyogo, Japan) in tight collaboration with Professor Yukihiro Ozaki and his team. Research interests comprise pharmaceutical, chemical, clinical and food analysis being conducted within several inter-/national (EU) third party funded projects and in tight collaboration with well-known industrial partners. Thereby, special emphasis is put onto the theoretical and technical development of NIRS.

Research leader Harald Martens

Dept. Engineering Cybernetics

NIRS in Big Data Cybernetics

Monday June 12, 08.30 – 09.00

The field of NIRS has for many years pioneered a culture and – methodology for combining many non-selective high-speed measurement channels by multivariate data modelling to attain rapid, relevant, reliable and interpretable predictions of chemical, physical and biological qualities. This chemometrics approach is now being successfully generalized to include other measuring principles and new application settings. Applied in continuous monitoring situations, e.g. in industrial processes or hyperspectral cameras, these high-dimensional instruments generate what I call Quantitative Big Data. For several hundred years since the Enlightenment, our civilization has become smarter and smarter by striving to understand the real world. Now we risk becoming dumber and dumber, using “deep learning” algorithms to find patterns and make predictions, without helping us to understand what these patterns are. The present lecture concerns how to use machine learning with a chemometric flavor instead, “keeping people in the loop”, e.g. in an advanced process control setting (what we at NTNU call “Big Data Cybernetics”). A new tool for rational multivariate data modelling of “ever-lasting” high-dimensional data streams, On-The-Fly-Processing (OTFP) was recently demonstrated by Vitale et al. 2017 ( https://www.researchgate.net/publication/311482705_On-the-fly_processing_of_continuous_high-dimensional_data_streams). In my lecture, I shall explain and demonstrate an industrial implementation of this OTFP method, as applied to hyperspectral environmental drone data and other types of “overwhelming”, continuous vis/NIR data streams. The raw spectra quickly generate files that are far too big to be read into a PC memory for conventional chemometric modelling. But the OTFP software continuously reads new spectra and develops a bilinear data model – or several such models – of the data in order to accumulate all systematic spectral variation patterns while deleting the random noise. The system gives rational error warnings and good graphical overviews of the essence in the data, and provides quantitative calibration/prediction as well as temporal and spatial forecasting. I shall also demonstrate a new tool for “de-shadowing” of hyperspectral aerial images, identifying and separating light source variations (sunny/ shaded /cloudy) from ground property variations (geology, chemistry, biology.

Biography: Harald Martens took his MSc in 1971 in industrial biochemistry, and worked for some years on protein chemistry in food science at the Norwegian Food Research Institute before turning to multivariate soft data modelling (psychometrics / chemometrics / sensometrics). He has published two books and several hundred research papers in the field, particularly concerning NIRS and other types of bio-spectroscopy, now cited more than 17 thousand times (see https://scholar.google.com/citations?user=60HNWsYAAAAJ&hl=da). Over the years, he has worked in a number of universities and companies on several continents. He is presently research leader in the “dynametrics” start-up -company Idletechs AS and external professor at the Department of Engineering Cybernetics at the Norwegian U. of Science and Technology (NTNU), Trondheim Norway.

Pricipal Scientist Erik Skibsted

Novo Nordisk

Current and Future Applications of Near-infrared in Pharmaceutical and Biopharmaceutical Industry

Tuesday June 13, 08.30 – 09.00

Near-infrared spectroscopy is one of the most robust and yet informative analytical technologies used in the pharmaceutical and biopharmaceutical industry. Regulatory agencies support the use of NIR by guidelines and everyday are researchers coming up with new applications, which are used for raw material characterization, formulation and process development studies, active process monitoring and control and final drug product quality testing. In this presentation will the speaker provide an overview of applications and discuss the future of NIR in the industry

 Biography: Erik Skibsted has a PhD degree from the University of Amsterdam where he did research in the use of near-infrared spectroscopy in solid dosage form manufacturing. He has 17 years of experience with the use of NIR for various applications within pharmaceutical and biopharmaceutical industry

Frans W.J. van den Berg

University of Copenhagen

NIRS in process control … “time is on my side”(1)

Wednesday June 14, 08.30 – 09.00

Automated process control systems are at the heart of contemporary production and near infrared spectroscopy (NIRS) was – together with gas-chromatography – the early enabler in the move from inferential control (process regulation based on indirect signals such as temperatures, flows and pressures) to compositional control. The latter does not only provide direct, dynamic insight into the chemical and physical state of the product2, it can also greatly simplify the control strategy.

Although the number of NIRS implementations in industry is growing steadily3, there is still plenty of interesting puzzles to solve in business and academia. This contribution takes a look at some fundamentals of process control and the utilization of NIRS. It will address the expertise gap (still present) between blue color control engineers and white coat analytical chemists. The potential and limitations of NIRS in process dynamics will be discussed, and based on these application areas with growth potential will be identified.

1 The Rolling Stones (1964)

2 A. Klimkiewicz, P. Paasch Mortensen, Ch.B. Zachariassen, F.W.J. van den Berg ‘Monitoring an enzyme purification process using on-line and in-line NIR measurements’ Chemometrics and Intelligent Laboratory Systems 132(2014)30-38

3 F.W.J. van den Berg, Ch.B. Lyndgaard, K.M. Sørensen, Søren B. Engelsen ‘Process Analytical Technology in the Food Industry’ Trends in Food Science and Technology 31(2013)27-35

Biography: Frans van den Berg is associate professor in Advanced Chemometrics in Food Technology at the University of Copenhagen, Denmark. He has a civil engineering degree in Laboratory Information and Automation, and MSc in Analytical Chemistry and a PhD in Process Analysis and Chemometrics. His interests are in the area of real-time process measurements, and statistical/data-driven monitoring, control and optimization. In his daily research activities he cooperates closely with the Danish and international dairy/food and biotech industries, and is a regular teacher in industrial short courses. At university level he is main responsible for the MSc courses Design of Experiments and Optimization, Food Process Equipment and Process Analytical Chemistry and Technology.

ICNIRS klodser