A portable electronic nose system for the identification of cannabis-based drugs

We report on a research aimed at evaluating the capacity of a simple, low-cost, portable electronic nose system based on commercially available metal oxide gas sensors to classify different types of drugs. Five drugs, namely cannabis buds, cannabis plants, hashish, snuff tobacco and tobacco leaves were employed. A dedicated real-time data acquisition system based on dynamic headspace sampling, a microcontroller and a laptop computer have been designed and constructed for this application. To demonstrate its discrimination capability, unsupervised and supervised classification models have been built and validated. Principal Component Analysis (PCA) of volatile profiles revealed five distinct groups corresponding to the five different drugs analyzed. This was further confirmed by a multivariate analysis of variance (MANOVA) test. Support Vectors Machines (SVMs) were applied to build a classifier and reached a 98.5% success rate in the recognition of the different drugs analyzed. This work demonstrates for the first time that the electronic nose technology could be successfully applied to the identification of illegal drugs.