Pattern recognition techniques applied for the detection of amphetamines based on infrared laser spectroscopy

We are presenting an exploratory analysis assessing the feasibility of detecting amphetamines based on their absorptions in the 1550-1330 cm^-1 spectral window, in which the tested External Cavity (EC) Quantum Cascade Laser (QCL) is emitting. The Principal Components Analysis indicates that amphetamines can be detected efficiently, this pattern cognition method generating models distinguishing these illicit compounds according to their biological activity. The predictive power of the first three principal components has been evaluated based on estimated density distributions. The sensitivity and selectivity of the class identity assignment was assessed by using a hierarchical clustering algorithm, i.e. agglomerative clustering. The results indicate that the evaluated EC-QCL is an adequate source for advanced portable sensors.