Laser-Induced Breakdown Spectroscopy Combined with Artificial Neural Network for Pre-carbonization Detection in Laserosteotomy

Abstract

To obtain efficient laser ablation in bone, dehydration, early carbonization and carbonization need to be avoided. Achieving this can only be provided by using an automated control of the ablation laser and irrigation system. As a preliminary study, we demonstrated a laser-induced breakdown spectroscopy based early carbonization detection system by analyzing carbonized bone tissues. Carbonization of bone samples was generated in a controlled way, by applying different number of Er:YAG pulses (0–25) at different locations on bone sample. To detect number of applied pulses, leading to the detection of carbonization level, we used a feed-forward Artificial Neural Network (ANN) with multi-layer perceptron structure. The results of the ANN were compared with the actual label, and R-squared of 0.85, 0.88, 0.86, 0.83, and 0.84 (0.85 on average) were achieved.