A novel neural network-based data acquisition system targeting high-speed electrical impedance tomography systems

**Note:** This preprint has been accepted for publication in [ IEEE Transactions on Circuits and Systems--I: Regular Papers ]. The published version is available on IEEE Xplore: [DOI link]( https://ieeexplore.ieee.org/document/10607897 ). The paper presents a design for a high-speed data acquisition (DAQ) system for electrical impedance tomography (EIT). The proposed solution involves using a high-speed analog-to-digital converter (ADC) to digitize the analog signals corresponding to multiple pairs of electrodes within the same cycle in a time-multiplexed manner. The extracted samples are fed to an artificial neural network (ANN) to estimate the peak amplitudes of the signals for every channel, which are then used for image reconstruction. Various ANN models with customized loss functions were assessed, and the optimal model selection approach using the grid search technique is presented. Unlike other multi-frequency based techniques, the suggested approach does not require a multi-frequency current source, thus simplifying the data acquisition system by not requiring high-quality narrow-band pass-band filters for different frequencies. The proposed approach allows EIT systems to operate at a very high throughput that can exceed 2,800 frames per second for a 50 kHz excitation signal using 32 or more electrodes. Extensive experimental testing showed that peak estimation accuracy can be achieved with more than 98%, even for signals with 40 dB SNRs. The suggested approach has thus promising potential for EIT applications requiring high SNR and fast data acquisition.