Abstract: Fruit trees can often suffer from the frost damage at low temperatures. The fruit tree stems can be gradually frozen to disrupt the internal water transport after cork formation. Effective identification of frost damage is required for early protective measures to reduce the losses in the yield of fruit trees. Commonly-used techniques can be used to detect the internal freezing and thawing in trees, such as calorimetry, ultrasonic emission, dielectric , nuclear magnetic resonance imaging, and X-ray imaging. However, only electrical impedance tomography (EIT) can reflect the impedance distribution in the image form, thus revealing the internal changes during freezing and thawing of fruit tree. In this study, the electrical impedance imaging was proposed to identify the freezing and thawing state of fruit tree stem. A 16-electrode EIT system was designed to connect the hardware in a stacked form, such as the power supply, signal generation, voltage-current conversion, multiplexing, signal demodulation, control unit, and temperature acquisition. In power supply module, various voltage regulators and boost converters were employed to provide the different positive and negative power supplies. While the signal generation module was used to output the stable sinusoidal voltage signals with adjustable frequency. The constant current source was utilized to serve as the enhanced Howland current source. The multiplexing module was used to switch among 16 electrode channels. The signal demodulation module was used to obtain the imaging data, and then demodulate the amplitude and phase of the measurement signal. The accuracy of this process was dominated the imaging. Analog demodulation was then chosen for this design. The control unit module was acquired and then stored the raw voltage data from EIT measurement. The wireless connections were also established with the host computer system via serial communication and Bluetooth modules. Temperature acquisition was integrated into the system as an essential part of freezing and thawing experiments. The system software included a Windows-based host computer system developed in C#. Control and monitoring of the hardware system were realized to determine the experiment parameters, and then record the data. The open-source EIDORS package was also incorporated to solve the imaging forward and inverse problems. System evaluation tests were carried out to clarify the performance and measurement, including constant current source output impedance, channel consistency, temperature correction, imaging, and measurement range. The output impedance of the constant current source was also tested after evaluation. The simulation results showed that the output impedance at 0-100 kHz was in the MΩ level. The measured impedance failed to reach the simulated value with the increasing frequency, but it was close to the MΩ level in the low frequency range. Three datasets of U-shaped curve were consistent after simulation, indicating the better consistency of channel. Furthermore, the standard deviation of the output amplitude significantly decreased after temperature correction. Differential imaging experiments were carried out on the cylindrical bodies with circular, triangular, and square insulation. Better imaging was achieved in the circular bodies, with the slight distortion at the edges of triangular and square bodies, yet essentially matching the shapes. The better performance of imaging was obtained on the distribution of hollow gypsum moisture with a diameter of 3 cm. Since the water content was generally lower than that of gypsum for actual stem objects, the actual measurement was less superior to the simulated gypsum objects. Therefore, the measurement object was tentatively set to small stems below 3 cm in diameter. Laboratory experiments on fruit tree stems confirmed that the conductivity was attributed to the great variation in the ice water content during stem freezing and thawing. The temperature gradually decreased over time during freezing. Among them, a latent heat process where the temperature remained almost constant, often corresponded when the water inside the stem converting to ice. The average conductivity also gradually decreased over time. In thawing, the temperature gradually increased over time, with a latent heat process corresponded when ice inside the stem converting to water. The average conductivity also gradually increased over time. Specifically, both ice content and overall impedance increased during freezing, whereas, the conductivity decreased. Both the ice content and overall impedance decreased during thawing, whereas, the conductivity increased. The freezing and thawing of the stem were observed using time-difference imaging of moisture distribution. The regions with the lower conductivity was continuously expanded during freezing, corresponding to the decrease in liquid water content. Conversely, the regions with the lower conductivity was gradually shrunk during thawing, corresponding to the increase in water content within the stem. Both processes were consistent with the average conductivity. The impedance imaging system can provide the better experimental performance to identify the freezing and thawing state of stems with a diameter of 3 cm.