Table of contents

An accurate impedance model of a skull plays an important role in the simulation research on source localization of EEG and brain EIT (electrical impedance tomography), etc. On the basis of the large number of impedance and resistivity data obtained from our previous measurement on the live human skull, in this study we established the equivalent circuit models of six types of skull samples in the 30 Hz–3 MHz frequency range and analyzed the fitting performance of the models. The six types of skull samples are standard tri-layer, quasi-tri-layer, standard compact, quasi-compact, dentate suture and squamous suture. The results showed that the difference of the real part between the CPE (constant phase model) model and the measured data was less than 1% for all skull tissue types when the optimized characteristic parameters (ρ₀, ρ_∞, α and f_c) were adopted in the model. It is the first time studying the impedance model of different types of skulls, and it may provide accurate modeling of the skull to improve the accuracy of the related research on bioelectricity of the head and the biological effects of the electromagnetic field.

This paper describes an unsupervised signal processing method applied to three-channel unipolar electrograms recorded from human atria. These were obtained by epicardial wires sutured on the right and left atria after coronary artery bypass surgery. Atrial (A) and ventricular (V) activations had to be detected and identified on each channel, and gathered across the channels when belonging to the same global event. The algorithm was developed and optimized on a training set of 19 recordings of 5 min. It was assessed on twenty-seven 2 h recordings taken just before the onset of a prolonged atrial fibrillation for a total of 1593697 activations that were validated and classified as normal atrial or ventricular activations (A, V) and premature atrial or ventricular activations (PAA, PVA). 99.93% of the activations were detected, and amongst these, 99.89% of the A and 99.75% of the V activations were correctly labelled. In the subset of the 39705 PAA, 99.83% were detected and 99.3% were correctly classified as A. The false positive rate was 0.37%. In conclusion, a reliable fully automatic detection and classification algorithm was developed that can detect and discriminate A and V activations from atrial recordings. It can provide the time series needed to develop a monitoring system aiming to identify dynamic predictors of forthcoming cardiac events such as postoperative atrial fibrillation.

Polysomnography (PSG) is currently considered the gold standard for assessing sleep quality. However, the numerous sensors that must be attached to the subject can disturb sleep and limit monitoring to within hospitals and sleep clinics. If data could be obtained without such constraints, sleep monitoring would be more convenient and could be extended to ordinary homes. During rapid-eye-movement (REM) sleep, respiration rate and variability are known to be greater than in other sleep stages. Hence, we calculated the average rate and variability of respiration in an epoch (30 s) by applying appropriate smoothing algorithms. Increased and irregular respiratory patterns during REM sleep were extracted using adaptive and linear thresholds. When both parameters simultaneously showed higher values than the thresholds, the epochs were assumed to belong to REM sleep. Thermocouples and piezoelectric-type belts were used to acquire respiratory signals. Thirteen healthy adults and nine obstructive sleep apnea (OSA) patients participated in this study. Kappa statistics showed a substantial agreement (κ > 0.60) between the standard and respiration-based methods. One-way ANOVA analysis showed no significant difference between the techniques for total REM sleep. This approach can also be applied to the non-intrusive measurement of respiration signals, making it possible to automatically detect REM sleep without disturbing the subject.

In the clinical situation and in most research work, the analysis of respiratory system mechanics is limited to the estimation of single-value compliances during static or quasi-static conditions. In contrast, our SLICE method analyses intratidal nonlinearity under the dynamic conditions of mechanical ventilation by calculating compliance and resistance for six conjoined volume portions (slices) of the pressure–volume loop by multiple linear regression analysis. With the gliding-SLICE method we present a new approach to determine continuous intratidal nonlinear compliance. The performance of the gliding-SLICE method was tested both in computer simulations and in a physical model of the lung, both simulating different intratidal compliance profiles. Compared to the original SLICE method, the gliding-SLICE method resulted in smaller errors when calculating the compliance or pressure course (all p < 0.001) and in a significant reduction of the discontinuity error for compliance determination which was reduced from 12.7 ± 7.2 cmH₂O s L⁻¹ to 0.8 ± 0.3 cmH₂O s L⁻¹ (mathematical model) and from 7.2 ± 3.9 cmH₂O s L⁻¹ to 0.4 ± 0.2 cmH₂O s L⁻¹ (physical model) (all p < 0.001). We conclude that the new gliding-SLICE method allows detailed assessment of intratidal nonlinear respiratory system mechanics without discontinuity error.

Photoplethysmography variability (PPGV) is currently considered to be a good surrogate to heart rate variability (HRV) measurements using the time between two pulse waves instead of RR intervals. Nevertheless, the interchangeability between HRV and PPGV has never been evaluated in situations with severe alterations in the autonomic nervous system (ANS). We aimed to identify the conditions for a correct utilization of PPGV in evaluating the consequences of sympathetic stimulations. Nine subjects performed three tests: active orthostatic test, slow walk and moderate and exhaustive cycling exercises in the supine position. Pulse waves at the fingertip and RR intervals were recorded at the same time. We used correlations and the Bland and Altman method to compare and evaluate interchangeability between several HRV indices. Bland and Altman analysis highlighted small discrepancies between PPGV and HRV for all HRV indices in the supine position and for , , LF_peak and RMSSD in the upright position. During the slow walk, it was impossible to detect properly PPG peaks. We observed large differences between the two methods during the cycling exercise. In conclusion, PPGV can be used instead of HRV without reserve in the supine position but only for some HRV indices in the upright position and not during slow walk and cycling exercise.

Extracellular fluid volume (ECV) is studied infrequently. The zero-time distribution volume (Vd) generated in the slope-intercept technique for measuring the glomerular filtration rate (GFR) substantially overestimates ECV. The aim was to validate a new technique for measuring ECV from the slope-intercept approach. GFR and ECV were measured using Cr-51-EDTA and iohexol injected into opposite arms in 51 patients undergoing routine measurement of GFR and on 48 occasions in 20 healthy volunteers. Blood samples were obtained bilaterally 20, 40, 60, 120, 180 and 240 min post-injection and assayed for indicator injected contralaterally. Reference ECV (ECV6) was calculated from all six samples as the product of indicator transit time and multi-sample GFR. GFR/ECV was calculated as the rate constant of the exponential fitted to the last three samples (GFR/ECV3). Slope-intercept GFR was calculated from the last three samples using the slope-intercept technique (GFR3). ECV (ECV3) was calculated by dividing GFR3 by GFR/ECV3, having corrected both for their one-compartment assumptions. ECV6_EDTA correlated closely with ECV3_EDTA (ECV3_EDTA = 1.01.ECV6_EDTA–0.5 L; r = 0.97; n = 99), but less closely with Vd (Vd = 1.17.ECV6_EDTA + 0.7 L; r = 0.86). ECV6_iohexol correlated slightly better with ECV6 _EDTA (ECV6_EDTA = 0.81.ECV6_iohexol + 3.3 L; r = 0.86) than with ECV3_EDTA (ECV3_EDTA = 0.83.ECV6_iohexol + 2.9 L; r = 0.84) and had slightly narrower 95% limits of agreement (−3.82 and 2.82 L versus −3.90 to 3.43 L). In conclusion, ECV can be measured from three samples almost as accurately as ECV from multiple samples.

Electrocardiogram (ECG) is required during magnetic resonance (MR) examination for monitoring patients under anaesthesia or with heart diseases and for synchronizing image acquisition with heart activity (triggering). Accurate and fast QRS detection is therefore desirable, but this task is complicated by artefacts related to the complex MR environment (high magnetic field, radio-frequency pulses and fast switching magnetic gradients). Specific signal processing has been proposed, whether using specific MR QRS detectors or ECG denoising methods. Most state-of-the-art techniques use a connection to the MR system for achieving their task, which is a major drawback since access to the MR system is often restricted. This paper introduces a new method for on-line ECG signal enhancement, called ICARE, which takes advantage of using multi-lead ECG and does not require any connection to the MR system. It is based on independent component analysis (ICA) and applied in real time. This algorithm yields accurate QRS detection for efficient triggering.

Myoelectrical pattern classification is a crucial part in multi-functional prosthesis control. This paper investigates a discriminant Fourier-derived cepstrum (DFC) and feature-level post-processing (FLPP) to discriminate hand and wrist motions using the surface electromyographic signal. The Fourier-derived cepstrum takes advantage of the Fourier magnitude or sub-band power energy of signals directly and provides flexible use of spectral information changing with different motions. Appropriate cepstral coefficients are selected by a proposed separability criterion to construct DFC features. For the post-processing, FLPP which combines features from several analysis windows is used to improve the feature performance further. In this work, two classifiers (a linear discriminant classifier and quadratic discriminant classifier) without hyper-parameter optimization are employed to simplify the training procedure and avoid the possible bias of feature evaluation. Experimental results of the 11-motion problem show that the proposed DFC feature outperforms traditional features such as time-domain statistics and autoregressive-derived cepstrum in terms of the classification accuracy, and it is a promising method for the multi-functionality and high-accuracy control of myoelectric prostheses.

Localized impedance methods can provide useful approaches for assessing neuromuscular disease. The mechanism of these impedance changes remains, however, uncertain. In order to begin to understand the relation of muscle pathology to surface impedance values, 8 immature rats, 12 mature rats and 8 mature rats that had undergone sciatic crush were killed. Measurement was made on tissue from the gastrocnemius muscle from each animal in an impedance cell, and the conductivity and relative permittivity of the tissue were calculated in both the longitudinal and transverse directions for frequencies of 2 kHz to 1 MHz. In addition, quantitative histological analysis was performed on the tissue. Significant elevations in transverse conductivity and transverse relative permittivity were found with animal growth, but longitudinal values showed no difference. After sciatic crush, both transverse and longitudinal conductivity increased significantly, with no change in the relative permittivity in either direction. The frequency dependence of the values also changed after nerve injury. In the healthy animals, there was a strong linear relation between measured conductivity and relative permittivity with cell area, but not for the sciatic crush animals. These results provide a first step toward developing a comprehensive understanding of how the electrical properties of muscle alter in neuromuscular disease states.

In this study we have investigated the accuracy of an accelerometer sensor designed for the measurement of cardiac motion and automatic detection of motion abnormalities caused by myocardial ischaemia. The accelerometer, attached to the left ventricular wall, changed its orientation relative to the direction of gravity during the cardiac cycle. This caused a varying gravity component in the measured acceleration signal that introduced an error in the calculation of myocardial motion. Circumferential displacement, velocity and rotation of the left ventricular apical region were calculated from the measured acceleration signal. We developed a mathematical method to separate translational and gravitational acceleration components based on a priori assumptions of myocardial motion. The accuracy of the measured motion was investigated by comparison with known motion of a robot arm programmed to move like the heart wall. The accuracy was also investigated in an animal study. The sensor measurements were compared with simultaneously recorded motion from a robot arm attached next to the sensor on the heart and with measured motion by echocardiography and a video camera. The developed compensation method for the varying gravity component improved the accuracy of the calculated velocity and displacement traces, giving very good agreement with the reference methods.

This study provides an important contribution to the definition of the expiratory flow increase technique (EFIT). So far, no measuring means were suited to assess the manual EFIT performed on infants. The proposed method aims at objectively defining the EFIT based on the quantification of pertinent cognitive parameters used by physiotherapists when practicing. We designed and realized customized instrumented gloves endowed with pressure and displacement sensors, and the associated electronics and software. This new system is specific to the manoeuvre, to the user and innocuous for the patient. Data were collected and analysed on infants with bronchiolitis managed by an expert physiotherapist. The analysis presented is realized on a group of seven subjects (mean age: 6.1 months, SD: 1.1; mean chest circumference: 44.8 cm, SD: 1.9). The results are consistent with the physiotherapist's tactility. In spite of inevitable variability due to measurements on infants, repeatable quantitative data could be reported regarding the manoeuvre characteristics: the magnitudes of displacements do not exceed 10 mm on both hands; the movement of the thoracic hand is more vertical than the movement of the abdominal hand; the maximum applied pressure with the thoracic hand is about twice higher than with the abdominal hand; the thrust of the manual compression lasts (590 ± 62) ms. Inter-operators measurements are in progress in order to generalize these results.

The purpose of this study was to use laser displacement sensors to examine the cross-correlation of surface mechanomyographic (MMG) signals detected from the rectus femoris muscle in perpendicular and transverse axes during isometric muscle actions of the leg extensors. Ten healthy men (mean ± SD age = 22.1 ± 1.6 years) and ten healthy women (age = 24.4 ± 2.8 years) volunteered to perform submaximal to maximal isometric muscle actions of the dominant leg extensors. During each muscle action, two separate MMG signals were detected from the rectus femoris with laser displacement sensors. One MMG sensor was oriented in an axis that was perpendicular (PERP) to the muscle surface, and the second sensor was oriented in an axis that was transverse (TRAN) to the muscle surface. For each subject and force level, the MMG signals from the PERP and TRAN sensors were cross-correlated. The results showed maximum cross-correlation coefficients that ranged from R_x,y  = 0.273 to 0.989, but all subjects demonstrated at least one coefficient greater than 0.89. These findings showed a high level of association between the MMG signals detected in the perpendicular and transverse axes. Thus, it may not be necessary to detect MMG signals in multiple axes.

The PDF file provided contains web links to all articles in this volume.