Impedance measurements for demineralized tooth lesions assessment

This work deals with the design and development of a non invasive though rather effective solution to detect carious lesions. The main aim of the study is to develop a simple automatic approach for the assessment of the tooth demineralization, which represents the earlier stage of the dynamic carious process, by means of impedance spectroscopy. In particular, impedance measurements were carried out on 50 extracted human teeth. Teeth were demineralized in-vitro according to a validated protocol; then, a morphological analysis of the tooth surface was performed by scanning electron microscopy to confirm enamel demineralization. The proposed approach tries to take advantage of the change of impedance phase due to the demineralization process. Data analysis confirmed that the best frequency for discriminating between demineralized and non-demineralized teeth is about 15 Hz, and this parameter can be used for building up an automatic classifier based on Multilayer Perceptron (MLP) topology. Impedance data were processed by using a modified single neuron, which allows classifying demineralized and sound teeth with an error rate of about 7%, estimating therefore the presence of carious lesions in progress. The presented work can be considered as a feasibility study with a final goal to conceive a simple and low-cost measurement system to identify caries at an early stage