Gelatin-based artificial nose for disease detection
Volatile organic compounds (VOCs) are organic compounds that evaporate at room temperature and are produced by most organisms. Accumulating evidence indicates that alterations in VOCs are associated with several diseases including cancer and tuberculosis.
Novel biomaterials for electronic nose construction
Electronic noses (e-noses) is a technology that uses chemical gas sensors capable of detecting gases, VOCs, or odour and can be applied in many fields. Research in the field of gas-sensing materials is expanding rapidly. The key objective of the EU-funded SCENT project was to develop materials that alter their optical and electrical properties in the presence of VOCs. “These new materials are gels containing gelatin, as the one we use for cooking, that is prepared in a way that becomes air stable and encapsulates optical and electrical probes used for sensing,” explains project coordinator Cecília Roque. As the gel materials alter their properties in the presence of certain VOCs, they can be used as sensors that provide a signal when exposed to gaseous samples. These sensors were incorporated in tailor-made e-nose devices in a detection chamber that mimics our nasal cavity. The device also comprised a signal transduction system that mimics electrical impulses sent to our brain upon VOC-binding to the olfactory receptors in our nose. The SCENT team developed new gel-based materials consisting of unique biological and chemical components. They also generated gas-sensing materials from plant sources as an environmentally friendly solution.
AI to enhance pathogen identification through VOC
“The e-nose developed in the SCENT project works similarly to our sense of smell. It requires ‘training’ first, by exposing the sensors to known samples and collecting signals,” emphasises Roque. These signals are used to develop the machine learning algorithm, that can make predictions regarding the nature of a particular sample. Using this approach, researchers produced signature VOCs for the classification of different pathogens. Using a large set of data, they associated patterns of VOCs with microorganism pathogenicity. They discovered that 18 VOCs is a sufficient number to predict the identity of pathogens with 77 % accuracy and up to 100 % precision. Moreover, they generated sets of VOCs that can predict the presence of a pathogen in a sample with high accuracy and precision. The established pathogen-VOC database from distinct biological samples can serve as the foundation for the future clinical implementation of such gas sensors. Importantly, the classification algorithms can be further trained with experimental evidence to increase the sensitivity of detection.
Advancing clinical diagnostics
Current methods for microbial detection in the clinical setting take about 24-36 hours, while for slow-growing bacteria it may take up to a week. Given the prevalence of antimicrobial resistance, e-noses capable of detecting bacterial VOCs as infection biomarkers offer a faster and equally sensitive approach. According to Roque, “The SCENT toolbox can be further used for several applications in healthcare, security or food industry for example.” The team is currently looking at the technological validation of the e-nose for the non-invasive monitoring of bladder cancer patients, through the ERC proof-of-concept project ENSURE.
Keywords
SCENT, volatile organic compounds, VOC, e-nose, gel, machine learning, algorithm, bacterial infection