Analysis of Parameters of Bioelectrical Impedance for Motion Discrimination

Abstract

We have proposed biodynamic analysis using equivalent series resistance of bioelectrical impedance at 50 kHz. This method uses a human body itself as a part of the sensor, and the change of bioelectrical resistance is measured in human motion. The analysis of bioelectrical impedance with some independent parameters is used for accurate evaluation or discrimination of human motion. This paper shows results of motion discrimination of wrist joint using four parameters of forearm electrical impedance: Z0 (resistance when frequency is 0 Hz), Z∞ (resistance when frequency reaches infinity), fm (frequency which gives maximum reactance of impedance) and β (ratio of central angle of circular arc to π and degree of distribution of relaxation time). In order to analyze impedance parameters, we have developed a measurement system for bioelectrical impedance with 10 frequency points and time resolution of 1 ms. The frequency characteristics of forearm electrical impedance were measured in the following six movements: palmar flexion, dorsi flexion, ulnar deviation and radial deviation of wrist joint, and pronation and supination of forearm. Each flexion angle was 60 deg except ulnar/radial deviation whose angle was 20 deg. Discriminant analysis was used to determine which change ratios of four parameters of forearm electrical impedance with each movement discriminate between six motions. As results of a subject with 60 data, hitting ratio was very high and 95.0%. Two supinations and on radial deviation were mistaken. There were shorter Mahalanobis distances between centroid of supination groups and centroid of ulnar/radial deviation groups. When discriminant function calculated using data from a subject was used to determine motions of other subject, the hitting ratio was 70.0%. This was caused by differences of impedance change between subjects in even same motion. This analysis using parameters of electrical impedance could be motion discrimination.