Bioimpedance assessment method based on back propagation neural network for irreversible electroporation of liver
Chengjiang Wang1,2, Yuchi Zhang1, Fulai Lin1,2
1National Local Joint Engineering Research Center for Precision Surgery & Regenerative Medicine, The First Affiliated Hospital of Xi'an Jiaotong University, Xi'an, 710061, China.
Abstract:
The safety and efficacy of irreversible electroporation (IRE) in tumor therapy has been validated over many years by clinical application. An in-depth study, however, is required to assess the degree of ablation during the clinical dissemination of the treatment. In this study, we propose and validate a method to evaluate the degree of IRE by measuring the impedance spectra of tissues before and after pulsed electric field treatment. IRE with varying parameters was applied to the liver tissue of mice to achieve varying degrees of ablation. Subsequently, the impedance spectra of the biological tissue were measured using an impedance analyzer at different time points before and after ablation, and the equivalent circuit method was used to quantify the results for analysis. We established a neural network model to investigate the relationship between the impedance after ablation and steady-state impedance after 72 h. Using ablation data of 55 mouse livers as training data samples and 5-fold cross-validation, the model predicted the equivalent circuit parameters after 72 h based on the equivalent circuit parameters of the tissues after 30 min of ablation. The model yielded acceptable prediction results with a root mean square error (RMSE) of 7.33, mean absolute percentage error (MAPE) of 8.62%, and coefficient of determination (R2) of 0.82. To explore the relationship between impedance changes and the degree of ablation at the steady state, an approximately exponential relationship between the relative changes in equivalent resistance of the extracellular fluid and the degree of ablation after 72 h was determined by performing ablation area measurements on the hematoxylin-eosin staining results of the samples under different impedance changes. This study demonstrate that the back propagation (BP) neural network can predict the steady-state impedance values after ablation within a short time and assess the degree of ablation based on the changes in impedance.
