Multifrequency EIT Calibration System Using Cucumber Samples

Multifrequency Electrical Impedance Tomography (MfEIT) is a safe, non-invasive, radiation-free technique that visualizes conductivity distribution within an object. Vegetables and animal tissues are often used to calibrate and validate MfEIT systems. This paper presents the calibration of image reconstruction algorithms using eight cucumber samples. An academic MfEIT model obtained edge voltage data at frequencies of 10k, 50k, 100k, and 200 kHz and gains of 5, 10, 60, 150, 200, and 250. Four algorithms were selected for image reconstruction: Graz Consensus for Electrical Impedance Tomography (GREIT); One-Step Gauss-Newton (OSGN); Total Variation Primal-Dual Interior Point Method (TV-PDIPM); and Gauss-Newton Absolute Reconstruction Method (GN-ARM). This work explores key factors influencing image quality, including gain, frequency, hyperparameter values, and reconstruction algorithms. Hyperparameters in algorithms play an important role in fine-tuning image reconstruction models, directly influencing the accuracy, stability, and computational efficiency of the obtained solutions. For this reason, they were systematically varied and tested in this work, aiming to identify the optimal configurations that ensure the best performance of the algorithms in different reconstruction scenarios. The cucumber sample at 250 gain and 100 kHz with GN-ARM produced the best image.