Computational Analysis of Electrical Impedance Spectroscopy for Margin Tissue Detection in Laparoscopic Liver Resection

Abstract

Margin tissue detection during intraoperative laparoscopic liver resection (LLR) is required to prevent tumor recurrence and reduce the likelihood of further surgery. This study proposes an electrical impedance spectroscopy (EIS) method for margin tissue detection in LLR to determine the boundary interface of normal and cancerous tissue. The proposed method of this study has three objectives: (1) designing the electrode array configuration to collect multiple EIS impedance measurements, (2) implementing the Feedforward Neural Network (FNN) to classify the orientation of margin tissue relative to the electrode array by using time-difference impedance indexes, and (4) governing the inflection point method based on impedance indexes to detect the margin tissue location. The proposed method is evaluated by a 3D numerical simulation of liver tissue composed of cancerous lumps with Iac = 1 mA alternating injection current at frequencies: lf = 1 kHz and hf = 100 kHz. The electrode array is composed of 16 electrode pairs each for injection current and voltage measurements. The variation of margin tissue orientation relative to the electrode array direction was considered to occur in unidirectional, perpendicular, and diagonal direction with noise variations (Signal-to-Noise-Ratio: 50 to 90 dB). The FNN trained on 2,400 data points achieves True Positive Rate (TPR) value as 90.2%, 99.4%, and 96.6% for diagonal, perpendicular, and unidirectional respectively in margin tissue orientation classification, while the inflection point method detects margin tissue location with 75% location at the unidirectional orientation (y-axis).

Keywords

• laparoscopy liver resection
• electrical impedance spectroscopy
• machine learning algorithm
• time-difference impedance indexes
• margin tissue detection