Bioimpedance spectroscopy measurement method based on multisine excitation and integer-period undersampling

Abstract

Bioimpedance spectroscopy (BIS) is a detection technology that uses the bioimpedance characteristics of human tissues and their changes to analyze their physiological and pathological status, and is widely used in clinical and scientific research applications. Traditional BIS measurement must satisfy the Nyquist sampling theorem so as to ensure that the measurement signal has no frequency aliasing, but at the same time the sampling frequency and the number of sampling points will be increased, which will increase the computation and hardware costs. This paper proposes a novel BIS measurement method based on multisine excitation and integer-period undersampling (IPUS) technology. Firstly, the multisine-based IPUS theory is deduced, and the BIS measurement principle based on multisine excitation and IPUS technology is introduced. Secondly, a BIS measurement system based on a field-programmable gate array + analog-to-digital converter + digital-to-analog converter architecture is designed, and multisine excitation with 32 pseudo-logarithmically distributed frequency components in the range of 2–997 kHz is generated. Comparative BIS measurement experiments on three RC three-element models are carried out under the Nyquist sampling condition (sampling frequency fs = 2.56 MHz) and under the IPUS condition (fs = 512 kHz), respectively. Experimental results show that the mean amplitude error of BIS measurement under the Nyquist sampling condition is 0.80% (±1.19% SD), while the mean amplitude error under the IPUS condition is 1.02% (±1.13% SD). Moreover, the signal-to-noise ratio (SNR z ) is calculated in 40 repeated BIS measurements, where the mean SNR z is 63.60 dB under the IPUS condition, similar to the value of 62.77 dB under the Nyquist sampling condition. The proposed multisine-based IPUS theory and its implementation method in this paper can complete a BIS measurement with only one fundamental period, and the sampling frequency and sampling point requirements are lower than for Nyquist sampling, laying a theoretical and technical foundation for a BIS measurement system with reduced hardware and computation requirements.