Abstract:
Hepatocellular carcinoma has been the leading cause of death in recent centuries and with the advent of newer technologies, several thermal and cryo-ablation techniques have been introduced in the recent past. In this regard, microwave ablation has developed into a promising method for thermal ablation technique. However, due to clinical obligations, in-vivo analysis is not feasible and ex-vivo analysis is inaccurate due to changes in the electrical and thermal properties of the tissue. Therefore, in this study, temperature-dependent permittivity, electrical conductivity, and thermal conductivity along with phase change effect due to temperature reaching above 100°C are incorporated using finite element method model. Further, using an intertwined normal mode helical antenna ablation probe, a change in resonant frequency (Δf) and reflection coefficient (ΔS11 ) from the actual value (antenna parameter in the air at 5 GHz) is modeled using second-order polynomial curve fitting to predict the surrounding permittivity in the range of 30-70. A maximum deviation of 0.8 value in permittivity from the actual value is observed. However, to obtain a generalized methodology, XG Boost and CAT Boost algorithms are used. Further, since ablation diameter plays a crucial role in achieving optimal tumor ablation, an artificial neural network (ANN) algorithm with three different optimizers is incorporated to predict ablation diameter using five critical parameters. Such an ANN algorithm which can predict the transversal and axial ablation zone may provide optimal ablation outcomes.
摘要:
肝细胞癌(HCC)已成为死亡的主要原因在最近几个世纪,随着新技术的出现,最近引入了几种热和冷冻消融技术。在这方面,微波消融(MWA)已发展成为一种有前途的热消融技术。然而,由于临床义务,体内分析是不可行的,并且由于组织的电特性和热特性的变化,离体分析是不准确的。因此,在这项研究中,与温度相关的介电常数,使用FEM(有限元方法)模型将电导率和热导率以及由于温度达到100°C以上而产生的相变效应结合起来。Further,使用iNMHA(交织正常模式螺旋天线)消融探头,使用二阶多项式曲线拟合对谐振频率(Δf)和反射系数(ΔS11)相对于实际值(5GHz空气中的天线参数)的变化进行建模,以预测30-70范围内的周围介电常数。观察到介电常数与实际值的最大偏差为0.8。然而,为了获得广义的方法论,使用XGBoost和CATBoost算法。Further,由于消融直径在实现最佳肿瘤消融中起着至关重要的作用,结合了具有三个不同优化器的ANN(人工神经网络)算法,以使用五个关键参数来预测消融直径。可以预测横向和轴向消融区域的这种ANN算法可以提供最佳消融结果。本文受版权保护。保留所有权利。