Essential and Natural Oils
The Sagitto system is ideal for the measurement of essential oils, and natural oils used for both cosmetic and food products. Accurate machine learning models can be built to rapidly quantify specific components in oils, using reference data created from conventional chemistry methods. In addition, samples can also be analysed for anomalies - instances where an individual sample differs from usual samples of that type.
Manuka flowers and leaves © wikipedia user gerald.w Copyrighted free use
This model measures the percentage of Manuka beta triketones in Manuka oil, using near-infrared (NIR) spectroscopy (900nm-1700nm). Oil extracted from Manuka plants harvested from the East Cape of New Zealand is high in natural β-triketones. The Sagitto system provides NZ Manuka Bioactives with a rapid and accurate way of grading their oils, and has become an integral part of ensuring that their oils comply with the MβTK™ grading system.
Oil extracted from Kanuka leaves is equally valuable. Since Manuka (Leptospermum scoparium) and Kanuka (Kunzea ericoides) plants are very similar in appearance, there is a risk that inexperienced staff involved in the harvesting process might accidently mix the two plant species. The Sagitto system provides NZ Manuka Bioactives with a valuable quality control mechanism, to ensure that their Manuka and Kanuka oils are pure.
There are many different Lavandula cultivars - each with its own fragrance profile and chemical composition. These differences are, of course, evident when the oils are analysed using gas chromatography. They are also reflected in their NIR absorbance spectra.
Sagitto uses data science to help plant breeders to rapidly screen cultivars in their search to maximise yield and/or develop new desirable combinations of traits.
By applying modern data science techniques to data from inexpensive and field-portable instruments, we can greatly speed up the discovery cycle since expensive and time consuming laboratory techniques only need to be applied to favourable candidates.
We can also analyse changes in spectra from an individual batch, to predict its optimum maturity date - just as the wine industry monitors the maturation of wines.
|The NIR spectra of an essential oil, such as lavender oil, is like a finger print. It is unique for each oil sample, at a particular time.||
|There are well-established data science techniques that can help us to visualise spectral data in two dimensions. However accurate prediction models use more sophisticated data science techniques to simultaneously examine hundreds of dimensions.||
|Here is an example report produced by the Sagitto system, using a machine learning model that classifies lavender oil by cultivar.||Example Report|
Harvesting Lavandula angustifolia 'Avice Hill' at Snowy River Lavender
This model measures the percentage (w/w) of squalene in shark liver oil, using near-infrared (NIR) spectroscopy (900nm-1700nm). The samples of shark liver oil used in this model had been previously analysed for squalene content by scientists at Plant and Food Research with results published in the paper Rapid Quantitative Determination of Squalene in Shark Liver Oils by Raman and IR Spectroscopy. Sagitto is grateful for the support of Plant and Food Research staff, who not only provided the samples but also the high quality reference data which allowed us to build a very accurate predictive model.