Feasibility of Computer-assisted Recognition of Different Dental Hard Tissues

Abstract

For pattern recognition, fuzzy set theory has been proven to be highly useful. The aim of the present investigation was to combine the fuzzy set with an ultrasonic scaler, to test its suitability for automatic detection of different tooth substances. An experienced operator placed the tip of a piezoceramic ultrasonic scaler on a tooth, thereby inducing oscillations in the contact area around the ultrasonic tip. Each surface showed a characteristic oscillatory behavior in the immediate vicinity of the tip. The oscillations were then re-transmitted to the scaler tip and recorded by the measurement of variations in current and voltage. Because the ultrasonic scaler is driven by piezoceramics, it can be used as both an oscillatory excitation and a sensor system. The data were processed with Fast Fourier transformation and analyzed by means of a fuzzy pattern recognition algorithm. Re-classification of the different measurements was carried out by the experienced operator's assessment. With a combination of six features (frequencies), re-classification was correct for 99% of all surfaces. The diagnostic reliability of the system was tested by the assessment of 50 teeth for which no learning data had previously been recorded. The unknown samples were correctly classified to 100%. The excellent results of these experiments suggest promising possibilities for the implementation of new diagnostic and therapeutic instruments in periodontology practice.