ORIGINALITY AND INNOVATION

Colour has been used by humans to judge Nature since the dawn of time. Our eyesight can pick up subtle colour differences and interpret what they mean. Colour can tell us whether a fruit is ripe or not, a mushroom is poisonous, the colour of your child’s face can indicate illness. Fluorescent microscopes have been the workhorses in biomedical laboratories since mid 20th century. It is quite surprising that, prior to our work, nobody has really put colour detection and microscopes together in a way that makes a genuine and novel difference for biomedical science.

Today’s modern day information technology and the prevalence of light emitting diodes (LEDs) for illumination provided the two key ingredients to our success in quantifying colour by hyperspectral imaging. After adding my expertise in chemistry and my desire to engage with biomedical research to the mix, the resulting combination provided us with interdisciplinary expertise essential to our success in pioneering hyperspectral imaging as a uniquely powerful cell and tissue characterisation method.

​

Hyperspectral imaging uses colour and spatial image information for detection and classification. We were the first to take the hyperspectral approach to the fields of medicine and cell biology. In our approach, we obtain fluorescence images of live cells/tissues at a number of selected excitation wavelengths, capturing their emission at multiple specified, (longer) wavelength ranges. This accurately quantifies their (fluorescence) colour. We analyse micrographs of cell populations using custom-developed software to gather information about hundreds of quantitative features including cell size, shape, brightness at each wavelength and texture etc. This dataset provides richly detailed information about living cells. Using this technique, we can non-invasively determine cell biochemistry. We are able to test hypotheses about similarity (or otherwise) of cell distributions and about the effects of chemical interventions, such as drug treatment. Putting it simply, we can look at the colour of patient’s cells and tell if the patient is sick or healthy, and how they'll respond to treatment.

​

First layer of innovation: our hardware. We took a standard fluorescent microscope and replaced its existing mercury lamp with a custom-made light source comprising inexpensive, multiple LEDs of different colours. We built a digital controller so that these LEDs could be switched on and off individually or collectively as needed.  And we added an off-the-shelf scientific image sensor to our digital controller. With this setup we were able to rapidly image our biological samples in about twenty colour channels, taking snapshots in each of those channel colours. With many LEDs and the corresponding channels we were able to document the colour of our cells and tissues extremely accurately. Our next milestone was to determine how this information can be best analysed. This is where chemistry and biology becomes critical. We have documented the fluorescent colour of our biological cells, so what exactly does it tell us?

​

Second layer of innovation: linking fluorescent colour with biochemistry of cells and tissues. Cells and tissues contain naturally occurring fluorescent molecules which provide signatures of key biological processes. These fluorescent molecules include nicotinamide adenine dinucleotide (NADH), nicotinamide adenine dinucleotide phosphate (NADPH), flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN), retinoids including a compound called A2E, cytochrome C, tryptophan plus many others. But when researchers look at cells they observe fluorescent colour with all of these signatures mixed together. We have developed mathematical algorithms to unmix the fluorescent colour into the individual components, which is the mathematical equivalent of taking a can of paint and separating out the many individual pigments that have been mixed together to create nuances of colour. But we have taken this a step further. Because we document cell images, we are able to extract the content maps of these fluorophores. This is like revealing the percentage content of each pigment used in every single spot in a painting. We related this information to cell biochemistry which is critically important as it provides the connection to the whole body of knowledge within the biology discipline. However, our methodology goes far beyond standard biochemistry because most biochemical assays require hundreds and thousands of cells while we work at the single cell level. An alternative approach to ours requires adulteration of cells with staining chemicals which limits in-vivo and medical applications. In this way hyperspectral imaging makes it possible to non-invasively extract rich, biologically relevant and quantitative information from native fluorescence of cells and tissues. Information which is otherwise inaccessible.

 

PUBLIC IMPACT

This pioneering hyperspectral colour-centric analytical approach will have significant and widespread translational outcomes and public impact. We reported the discovery of six separate major diagnostically relevant outcomes relating to (a) cancer; (b) neurodegeneration, (c) diabetes, (d) embryo quality, (f) stem cell subpopulations and (g) cell surface biomarker detection. This discovery can be used as novel non-invasive diagnostic point-of-care methodologies.

Our earlier research in the hyperspectral colour analysis has also discovered that we can detect mitochondrial dysfunction in neuronal cells. Neurodegeneration is a critical component of many eye diseases. As the retina is particularly easy to access microscopically, we developed a hyperspectral adaptation of an ophthalmological fundus camera to quantify retinal cell colour. Our technology uses small amounts of light, within safe exposure limits, enabling rapid translation to ophthalmology. We are already collaborating with several ophthalmologists to help realise the impact of hyperspectral imaging in eye disease.

​

POTENTIAL FOR YOUR FUTURE RESEARCH APPLICATIONS

As colour is such a universal and subtle biological characteristic, the hyperspectral technology has potential impact across the entire field of life sciences and medicine. As key cellular fluorophores are related to metabolic activity, using hyperspectral imaging we can non-invasively measure subtle physiological features such as the metabolic rate. Thus, hyperspectral imaging may help to better understand a broad range of diseases affecting metabolism including neuronal degeneration, cancer and diabetes.

​

Differences between cancer and non-cancer tissue are particularly pronounced, and hyperspectral imaging offers scope to be able to be used during surgery to help identify cancer margins. Our current project in the area of motor neuron disease is aiming to establish its early diagnostics. In the area of diabetes, our current project looks at colour characteristics of cells shed by the kidneys into the urinary tract, enabling the diagnosis of diabetic degeneration of kidneys. In the area of reproductive medicine we are working towards a commercial colour-based methodology to determine embryo quality. In the area of neuroscience, we are non-invasively detecting cytokine secretion, of key significance in neuroimmunology, with a view to detect objective colour biomarkers of chronic pain. This is significant, because currently, there is no objective measure to quantify pain. In regenerative medicine, we have the capacity to non-invasively identify stem cells in a mixed population, and we can track progress of such cells towards differentiation more accurately than with standard invasive labelling. We can also identify stem cell subpopulations with optimised properties, such as those that show rapid differentiation potential. Non-invasive identification of optimised cell populations is critical in the area of regenerative medicine and commercial cell technologies, a major area of research and commercial activity these days. We are able to use colour features as a proxy for standard cell biomarkers such as CD90, and we are able to use other colour features to non-invasively measure the level of reactive oxygen species. This is critical in ophthalmology as well as in monitoring of photodynamic therapy.  We are involved in a high-content / high-throughput drug screening project, where the ability to detect the effect of drugs without external labelling will greatly accelerate speed and lower cost of pharma research. There is scope to use this method in microbiology, to identify microbial species. The list of applications for this innovation is practically endless.

​

Commercial applications beyond health, in industrial and environmental monitoring will be the focus of our future work. We foresee major applications for the food industry, such as determining the quality of fresh and processed foods; in industrial processes to identify microbial contamination in ethanol production, for example, using image features we were able to identify percentage bacterial contamination of yeast with 5% accuracy. We discussed the possibility of a citizen science project where recreational sailors would monitor microbial content of the oceans, linking the data to climate patterns and climate change. What really helps is the affordability of our imaging system and the fact that it can be easily retrofitted. For example, non-invasive investigations of cell metabolism has been previously carried out by multiphoton microscopy and fluorescence lifetime imaging (FLIM) which require expensive infrastructure (in the range of $0.5M -1M). Our system costs less than $50K and it can be retrofitted onto existing fluorescence microscopes.

​

The potential applications for colour analysis are endless. But the best part is you can do it yourself! We have developed a version of our technology for use on a mobile phone. Download the app from cnbp.org.au, and you can do your own point-of-care screening, environmental assays, or engage in exciting citizen science projects.