LIBS for real-time monitoring of laser-inducedthermal damage in laserosteotomy: from dehydration to full carbonization

Laserosteotomes will only be clinically applied if they do not carbonize the tissue. This can only be ensured with an irrigation system with correctly tuned water and air flow rate. Well-designed irrigation helps avoiding thermal damage, increases ablation speed, and accelerates healing while too little water leads to bone carbonization, and too much water decreases the ablation efficiency [1]. State-of-the-art for determining the irrigation parameters is to apply water and air flow rates based on pre-performed experiments without monitoring the bone surface in real-time. The lack of real-time monitoring of the laser-induced thermal damage can thus result in carbonization, caused by either a malfunction in the irrigation system or by unexpected changes in the properties of the tissues from patient-to-patient (or even point-to-point due to an inhomogeneous structure). Authors recently demonstrated that LIBS can classify fully carbonized bone from not carbonized one [2]. While such a classification can avoid further carbonization, even localized carbonization on the surface of the bone can still prolong the healing process.There are several stages before full carbonization occurs, such as dehydration and early carbonization. The detection of these early signs makes prospective carbonization prediction possible. This study aims at finding a regression between all steps of laser-induced thermal damage. Therefore, the spectrum of localized plasma generated in the ablation spot, together with the number of shots sent to the bone without irrigation was recorded using an in-house fiber-coupled Echelle spectrometer. Later on, based on the recorded spectra, regression models were trained. The model used were regression tree(single tree and also an ensemble of several trees), support vector machine, and a convolutional neural network. Finally, we calculated the root mean square error of each model and applied the trained models to the data collected afterwards in real-time. References: [1] H. Abbasi, et al., "Effect of cooling water on ablation in Er:YAG laserosteotome of hardbone", Proc. SPIE 10453 (2017). [2] H. Abbasi, et al., "Laser-induced breakdown spectroscopy as a potential tool for autocarbonization detection in laserosteotomy", J. Biomed. Opt. 23 (2018). Acknowledgement The authors gratefully acknowledge funding of the Werner Siemens Foundation through the Minimally Invasive Robot-Assisted Computer-guided LaserosteotomE (MIRACLE) project.