Exploring the relationship between the dielectric properties and viability of human normal hepatic tissues from 10 Hz to 100 MHz based on grey relational analysis and BP neural network

Liver is an important parenchyma organ, and its tissue viability plays an important role in liver transplantation and liver ischemic injury assessment. Dielectric property is a useful biophysical feature that provides insights into the structure and composition of biological tissues. This work aims to establish the relationship between the dielectric properties and viability of human normal hepatic tissues and explore the possibility of evaluating tissue viability by using dielectric properties. First, data on dielectric properties and tissue viability (including cell morphology and enzyme indicators) were collected from human liver tissues at 0.25-24 h after isolation. Grey relational analysis was conducted to select dielectric property and tissue viability indices that were highly correlated with prolonged ex vivo time as the inputs and outputs, respectively, of back-propagation (BP) neural network analysis. Finally, a BP neural network was developed with the Levenberg-Marquardt algorithm to explore the possibility of using dielectric properties as the basis for tissue viability evaluation. Results showed that the mean relative error for prediction was 2.40%, indicating that the model showed potential in forecasting liver tissue viability by applying dielectric properties.

肝脏是重要的实质器官，其组织活性在肝移植和肝脏缺血性损伤评估中起着重要作用。介电特性是一种有用的生物物理特征，可深入了解生物组织的结构和成分。本研究旨在建立人正常肝组织介电特性与活性之间的关系，并探索利用介电特性评估组织活性的可能性。首先，从分离后0.25 - 24小时的人肝组织中收集介电特性和组织活性（包括细胞形态和酶指标）的数据。进行灰色关联分析，分别选择与体外时间延长高度相关的介电特性和组织活性指标作为反向传播（BP）神经网络分析的输入和输出。最后，利用Levenberg - Marquardt算法构建了一个BP神经网络，以探索将介电特性作为组织活性评估基础的可能性。结果显示，预测的平均相对误差为2.40%，表明该模型在应用介电特性预测肝组织活性方面具有潜力。