Automatic detection of coagulation and carbonization in laser applications using machine learning techniques

Abstract

Laser-induced thermal treatment (LITT) technique is considered to be the most effective method for minimally invasive surgical procedures. However, the parameters of laser systems used in surgical operations need to be adjusted very precisely. Otherwise, irreversible situations may occur especially in the operation of sensitive organs such as the brain. In order to avoid such undesirable situations, artificial intelligence algorithms are used in the fields of engineering, aerospace and automotive, especially in medicine and biomedical. In this study, it was aimed to automatically detect Coagulation (Co) and Carbonization (Ca) states using a total of 140 data obtained as a result of LITT experimental applications (1070 nm and seven different power density values) on ex vivo liver tissue. In the first part of the study, which consists of two main stages, the data set was presented to seven different expert systems based on artificial intelligence (C4.5, KNN, ANN, RF, MLR, SMO-SVM and LWLDS), and automatically Co and Ca formations were determined. Among the proposed expert systems, the C4.5 algorithm showed the best performance with 100% accuracy and 0% error rate. In the second stage of the study, the same data set was given to Attribute Information Gain (AIG) ranking system, which was used to determine the minimum number of feature that would provide the maximum performance, and the gain values of the attributes were calculated. As a result of this application, the most effective feature was found to be Normalized Depth of Thermal Damage (NDeTD) with 100% accuracy rate. Thus, the purpose of using the AIG system has been successfully achieved. As a result, thanks to the expert systems presented within the scope of the study, Co and Ca formations were determined automatically with 100% success rate according to the required parameters without laser application on ex vivo liver tissue. The fact that the aim of the study was realized with the highest success rate is thought to be the first in terms of contributing to the literature. Through this study, as a result of the laser treatment to be applied on cancerous tissue, it can be predicted whether the tissue will suffer irreversible damage and necessary precautions can be taken beforehand.