Abstract: The most commonly used drug testing methods are based on the analysis of hair and urine using gas chromatography-mass spectrometry, liquid chromatography-mass spectrometry or immunoassay screening. These methods are time-consuming and partly expensive. One alternative method could be the application of an “electronic nose” (eNose). We have developed an eNose to detect directly on the human skin surface metabolic changes in the human body odor caused by cannabis consumption. Twenty cannabis-smoking and 20 tobacco-smoking volunteers were enrolled in this study. For the sensor signal data processing, two different methods were applied: Principle component analysis (PCA) with discriminant analysis, and the method of pattern recognition with subsequent support vector machines (SVM) processing. The PCA analysis achieved a correct classification of 70%, whereas the SVM obtained an accuracy of 92.5% (sensitivity 95%, specificity 90%) between cannabis-consuming volunteers and tobacco-smoking subjects. This study shows evidence that a low-cost, portable and fast-working eNose system could be useful for health protection, security agencies and for forensic investigations. The ability to analyze human body odor with an eNose opens up a wide field for diagnosing other drugs and also various diseases.
Keywords: electronic nose; principle component analysis (PCA); support vector machine (SVM); pattern recognition; human body odor
Export to BibTeX
MDPI and ACS Style
Voss, A.; Witt, K.; Kaschowitz, T.; Poitz, W.; Ebert, A.; Roser, P.; Bär, K.-J. Detecting Cannabis Use on the Human Skin Surface via an Electronic Nose System. Sensors 2014, 14, 13256-13272.
Voss A, Witt K, Kaschowitz T, Poitz W, Ebert A, Roser P, Bär K-J. Detecting Cannabis Use on the Human Skin Surface via an Electronic Nose System. Sensors. 2014; 14(7):13256-13272.
Voss, Andreas; Witt, Katharina; Kaschowitz, Tobias; Poitz, Wolf; Ebert, Andreas; Roser, Patrik; Bär, Karl-Jürgen. 2014. "Detecting Cannabis Use on the Human Skin Surface via an Electronic Nose System." Sensors 14, no. 7: 13256-13272.