On-line respiratory artefact removal via adaptive FIR filters in rheopneumographic measurement

Abstract

Impedance rheopneumography is a simple non-invasive technique that can reflect the vascular condition in the human pulmonary circulatory system. However, the much larger and almost in-banding respiratory artefact present has greatly restricted its usefulness to only respiratory patients of a less severe type, in view of the existing practice of requiring the subject to stop breathing momentarily during measurement. Conventional fixed or adaptive filtering cannot satisfactorily remove the artefact in view of the non-time stationary characteristic of the latter. In the paper, a fast adaptive FIR filter design method, which is based on the filter coefficient look-up table (CLT) concept, is presented as a solution. The CLT is constructed with the coefficients indexed to the cut-off frequency to separate the two components of the plethysmogram. An on-line fast Fourier transform is calculated to track the cut-off frequency. This filter can adaptively change its coefficients, not only for different subjects, but also for the same subject during long-term monitoring. Results show that this filter design is capable of providing an almost respiratory artefact-free signal for a majority of patients. The high speed of implementation also renders it a possibility for real-time monitoring applications.