Validity of the Adaptive Filter for Accurate Measurement of Cardiac Output in Impedance Cardiography

Modified scaled Fourier linear combiner in thoracic impedance cardiography

New filtering algorithms; a modified scaled Fourier linear combiner (mSFLC) and a modified ensemble averaging (mEA) are proposed to remove the phase distortion of the impedance waveform caused by SFLC. Performance of the filters is assessed from the shape of the filtered waveform and the estimation of systolic time intervals. It was found that the mSFLC and mEA could compensate for drawbacks of the SFLC and precisely reproduce the impedance waveforms; however, misinterpretation and attenuation in the cardiac variables may be caused when abrupt changes or irregularities occur in the source waveform.

Modified thoracic impedance plethysmography to monitor sleep apnea syndromes

BACKGROUND AND PURPOSE

In order to identify sleep disorders by thoracic impedance plethysmography (TIP), we propose several new techniques: the application of an adaptive filter, a scaled Fourier linear combiner (SFLC) to eliminate cardiac-derived fluctuation in the impedance waveform, and the use of heart rate variability (HRV) to ascertain whether the airflow is obstructed.

PATIENTS AND METHODS

Laboratory simulation experiments on four healthy individuals and actual overnight measurements on five patients with sleep disorders were carried out. Amplified thoracic impedance change (DeltaZ), ECG, a phonocardiograph, a pneumotachograph and a standard polysomnograph were recorded. The SFLC was applied to DeltaZ to selectively extract the cardiac-synchronous component (DeltaZ(CSC)), and the remainder of the waveform (DeltaZ(REM)) was low-pass filtered to estimate the waveform driven by respiration. The HRV was divided into respiratory synchronous (HRV(R)) and low frequency (HRV(L)) components.

RESULTS

The SFLC could drastically extract DeltaZ(CSC) from DeltaZ and thereby demonstrate a DeltaZ(REM) pattern quite similar to the flow-volume curve of the pneumotachograph. Central sleep apnea could be identified as the cessation of DeltaZ(REM) and concomitant attenuation of HRV(R). Obstructive sleep apnea could be identified as the maintenance of rhythmic but attenuated variations of DeltaZ(REM) accompanied by asynchronous fluctuation of HRV(R) against DeltaZ(REM). Central hypopnea could be identified as a normal but attenuated waveform in both DeltaZ(REM) and HRV(R). A large fluctuation in HRV(L) was observed during repetitive appearances of apnea/hypopnea in the nocturnal experiments.

CONCLUSION

The modified TIP together with HRV provides a superior tool for accurate and convenient definition of sleep apnea syndromes.

Beat-to-Beat Evaluation of Systolic Time Intervals during Bicycle Exercise Using Impedance Cardiography

In order to elucidate the beat-to-beat changes of the systolic time intervals (STI) during exercise, we proposed new techniques relating to an adaptive filter and detection algorithms for B- and X-points in the impedance cardiograph (ICG). Six male subjects underwent a ramp bicycle exercise up to maximum intensity during which an ECG, ICG and phonocardiogram (PCG) were continuously measured. Following the application of an adaptive filter, the scaled Fourier linear combiner (SFLC), to the first derivative (dZ/dt) of the base impedance (deltaZ) and PCG waveforms, the B- and X-points were automatically determined. For the B-point detection we used three criteria: the zero-crossing point (B(zero)), the 15% response point (B15%) of the negative peak of the dZ/dt (dZ/dt(min)) and a new algorithm (B(new)). The X-point was separately determined by using the ICG and PCG waveforms. It was found that the shape of the dZ/dt waveform directly affected the determination of the B- and X-points. The B-points determined using B(zero) and B15%, criteria were sometimes unstable caused by the location of a notch preceding the dZ/dt(min) compared to the B(new). The time difference between the X-points measured by the ICG and PCG was mostly within +/- 20 milliseconds but statistically significant. Although a wide variation was seen in R-R intervals, the STI were more stable. The relationships between HR and STI from rest to maximal exercise showed a gentle curvilinear relationship. It is suggested that the STI can be obtained precisely on a beat-to-beat basis by using the adaptive filter and detection algorithms for the inflection points of the ICG even during maximum exercise.