Abstract
OBJECTIVE: This study evaluates the potential of complex 3D Electrical Impedance Tomography (EIT) for intraoperative surgical margin assessment (SMA) using an ex vivo bovine model. METHODS: A custom electrode array was used to collect impedance data across multiple frequencies (100 Hz-1 MHz) from 57 tissue samples. An optimal pressure range was identified to ensure proper electrode contact with minimal tissue deformation (2 kPa to 5 kPa) and a saline calibration was used to minimize model-data mismatch. A novel best fit factor, β, was introduced to scale the reference data of difference EIT and initial guess for absolute EIT to eliminate inversions in permittivity. The accuracy of complex EIT reconstructions were investigated using pixel-based tissue classification. RESULTS: Experimental difference EIT, particularly for conductivity (σ) at 100 Hz, achieved high sensitivity (0.92) and specificity (0.90) with fast reconstruction speeds of 1.67 seconds. While absolute EIT performed slightly better in quantitative metrics for σ imaging (AUC = 0.90, Accuracy = 0.86), difference EIT was found to be more suitable for real-time applications due to its faster processing time. Simulations suggested that permittivity (ϵ) has sufficient contrast for classifying muscle and adipose tissue. However, experimental ϵ reconstructions exhibited lower performance, suggesting the need for hardware improvements. CONCLUSION: Difference EIT is the most promising method for real-time applications, balancing high accuracy with reconstruction speeds. Absolute EIT, while slightly more accurate, is less feasible due to longer processing times (150.6 seconds). SIGNIFICANCE: This study advances the potential for real-time intraoperative SMA using a novel complex EIT reconstruction method.