Methodology for binary detection analysis of inkjet-printed optical sensors for chemical detection

Abstract

We have developed inkjet-printed surface-enhanced Raman spectroscopy (SERS) sensors that are specifically designed to work with field portable Raman analyzers for the detection of chemical and biological agents (Tay et al. in J Raman Spectrosc, 2020). These SERS sensors are inkjet-printed on filter paper or fabric substrates. They are flexible and lightweight and provide point-of-sampling capabilities that other rigid planar SERS substrates lack. Conventionally, performance of a SERS substrate is characterized by its enhancement factor which is defined as the ratio of the normalized SERS to normal Raman intensity. This is an inadequate measure for a chemical/biological sensor. The performance of a chemical/biological sensor is best characterized by its sensitivity, probability of true positive detection, false positive rate, and its response time. The sensor using receiver operating characteristic (ROC) captures the performance trade-off between all these factors. Here, we outline the application of ROC analysis to the SERS sensors. This can be applied for all SERS sensors but we will focus its use with the printed SERS sensor. Specifically, the use of printed SERS sensor for the detection of fentanyl, a common opioid found in illicit drug, analyzed with ROC concept is demonstrated.