Zusammenfassung
In process analytics or environmental monitoring, thereal-timerecording of the composition of complex samples over a long periodof time presents a great challenge. Promising solutions are label-freetechniques such as surface plasmon resonance (SPR) spectroscopy. Theyare, however, often limited due to poor reversibility of analyte binding.In this work, we introduce how SPR imaging in combination ...
Zusammenfassung
In process analytics or environmental monitoring, thereal-timerecording of the composition of complex samples over a long periodof time presents a great challenge. Promising solutions are label-freetechniques such as surface plasmon resonance (SPR) spectroscopy. Theyare, however, often limited due to poor reversibility of analyte binding.In this work, we introduce how SPR imaging in combination with a semi-selectivefunctional surface and smart data analysis can identify small andchemically similar molecules. Our sensor uses individual functionalspots made from different ratios of graphene oxide and reduced grapheneoxide, which generate a unique signal pattern depending on the analytedue to different binding affinities. These patterns allow four purinebases to be distinguished after classification using a convolutionalneural network (CNN) at concentrations as low as 50 & mu;M. Thevalidation and test set classification accuracies were constant acrossmultiple measurements on multiple sensors using a standard CNN, whichpromises to serve as a future method for developing online sensorsin complex mixtures.
Altmetric