Classification of human tooth using laser-induced breakdown spectroscopy combined with machine learning

Identification of human sex from cadaver examination is crucial for various fields like forensic odontology and archaeology. Sex recognition can be very challenging if the cadaver is exposed to extreme temperature and/or pressure. As the hardest part of the body, tooth remains can be useful in such circumstances. In this report, we demonstrate a proof-of-concept experiment for rapid recognition and classification of human sexuality from their tooth using laser-induced breakdown spectroscopy (LIBS) combined with machine learning algorithms. LIBS experiment was performed on the enamel part of the tooth samples of males and females. LIBS spectra of male and female teeth have shown a higher degree of similarity owing to the same chemical structure. Three machine learning algorithms including principal component analysis, artificial neural network, and logistic regression were used for their identification. The higher identification rates enable the strong possibility of human sex recognition from their teeth LIBS spectra. Further, a judicious feature selection approach was implemented to significantly reduce the data size to achieve maximum accuracy with minimal time. Finally, Student's t-test was applied with a 95% confidence level to both the datasets to identify the most relevant features. Our findings demonstrate that the LIBS coupled with machine learning can be used as a productive and handy tool for fast and accurate recognition of human sex/gender from their teeth samples in forensic odontology and archaeology practice.