Piezoelectric Sorption Anesthetic Sensor

The piezoelectric sorption detection method is evaluated for use in construction of an inexpensive, rugged, compact, and fast sensor for measuring the concentration of inhalation anesthetics. 10 MHz quartz crystals were coated with various substances and their frequency shifts measured upon exposure to varying concentrations of halothane, enflurane, and nitrous oxide. For a silicone rubber coating, the response was found to be linear with concentration, with sufficient signal to allow resolution to approximately 0.05 volume percent for either halothane or enflurane, or to 4 percent for nitrous oxide. The devices were sensitive to halothane and enflurane in proportion to their anesthetic potencies, but were slightly less sensitive to equally potent concentrations of nitrous oxide. They were insensitive to CO2 in the physiologic range, but had a significant response to water vapor. The response time of the detector was primarily a function of washout of the sensing chamber. At a 200 ml/min sample flow rate, the time constant was approximately 100 ins. It is concluded that this detection method is suitable for a variety of applications in anesthesia, although, with the coatings presented here, control of or correction for sample humidity is required.     

Forecasting the unresponsiveness to verbal command on the basis of EEG frequency progression during anesthetic induction with propofol

The objective of this study is to model the association between the electroencephalogram (EEG) spectral features and the novel r scale representing the sedative effects of the propofol anesthetic drug. On the basis of the r scale, the unresponsiveness to the verbal command (LVC) is forecasted. EEG recordings are taken from a 16-patient study population undergoing propofol anesthetic induction. EEG was filtered into consecutive 4-Hz passbands up to 28 Hz. Of these time-series, the amplitude envelopes were extracted and used as input features to the first and the second-order polynomial multiple linear regression models. The values r epsilon [0.4, 1] were predicted with the R2 value of 0.775 with a cross validation. The LVC times were forecasted with the median error of 5%-7% or equivalently 10-13 s. In contrast, using the median of the measured LVC times of the training population as a forecast, the corresponding error was 12% or 26 s. The results suggest an acceptable correlation between the r scale and the EEG spectrum in the studied range. Moreover, the r values of an individual can be predicted using a population model. The suggested framework enables forecasting the LVC, which may open new possibilities for steering the drug administration.     

Design of a recognition system to predict movement duringanesthesia

The need for a reliable method of predicting movement during anesthesia has existed since the introduction of anesthesia. This paper proposes a recognition system, based on the autoregressive (AR) modeling and neural network analysis of the electroencephalograph (EEG) signals, to predict movement following surgical stimulation. The input to the neural network will be the AR parameters, the hemodynamic parameters blood pressure (BP) and heart rate (HR), and the anesthetic concentration in terms of the minimum alveolar concentration (MAC). The output will be the prediction of movement. Design of the system and results from the preliminary tests on dogs are presented here. The experiments were carried out on 13 dogs at different levels of halothane. Movement prediction was tested by monitoring the response to tail clamping, which is considered to be a supramaximal stimulus in dogs. The EEG data obtained prior to tail clamping was processed using a tenth-order AR model and the parameters obtained were used as input to a three-layer perceptron feedforward neural network. Using only AR parameters the network was able to correctly classify subsequent movement in 85% of the cases as compared to 65% when only hemodynamic parameters were used as the input to the network. When both the measures were combined, the recognition rate rose to greater than 92%. When the anesthetic concentration was added as an input the network could be considerably simplified without sacrificing classification accuracy. This recognition system shows the feasibility of using the EEG signals for movement during anesthesia     

Closed-Loop Control of an Anesthesia Delivery System: Development and Animal Testing

An anesthesia system which integrates closed-loop control of ventilation, oxygen, nitrous oxide, and anesthetic agent delivery into a closed breathing circuit is described. Breathing circuit volume is regulated by controlling the sum of oxygen and nitrous oxide flow, and oxygen concentration in the breathing circuit is regulated by controlling the ratio of oxygen flow to the sum of oxygen and nitrous oxide flow. End-tidal anesthetic agent concentration is regulated by controlling the agent delivery, and end-tidal carbon dioxide concentration is regulated by controlling the ventilation. After tuning the ventilation and anesthetic delivery controllers in preliminary trials in ten dogs, the system was tested and evaluated in five dogs. All control loops were stable and exhibited time responses to step changes in setpoint or external disturbances which were clinically acceptable. The system makes available the advantages of closed-system anesthesia without encumbering the anesthesiologist with the control tasks associated with the technique.     

基于相位延迟指数的脑功能网络及测谎研究

在脑认知科学领域,越来越多的研究开始专注于利用不同导联脑电信号之间的相互依赖关系来研究大脑整体认知功能.相位延迟指数可有效减少由容积导体引起的误差,该方法已被广泛应用.而基于图论的脑网络研究方法在测谎方面还少见报道.本文通过对30名（诚实和说谎）受试者的脑电信号进行网络拓扑分析,将网络参数作为判别指标,使用支持向量机对实验数据进行分类.研究发现,两组受试者的小世界指标表现出显著的统计学差异,且得到较高的测谎准确率,结果证明了利用相位延迟指数方法进行图论分析的测谎有效性.     

Instrumentation for Measuring Continuous Oxygen Consumption of Surgical Patients

A system has been developed for continuously monitoring the oxygen consumption (V02) of surgical patients. A replenishment technique is used whereby the oxygen removed by the patient from a closed rebreathing circuit is replaced. This is accomplished using a feedback-controlled pump to add oxygen at a rate necessary to maintain a constant inspired oxygen percent. A second feedback loop adds nitrous oxide to the circuit at a rate equal to the patient's nitrous oxide uptake rate, thus maintaining a constant circuit volume. These two feedback loops constitute a system to monitor oxygen and nitrous oxide uptake during nitrous oxide anesthesia. The instrument responds to a step change in V02 in 4 min. System accuracy has been shown to be ±6% in an in vivo comparison, the major error resulting from oxygen sensor or electronics dnift, leaks in the system, or changes in residual volume. The record of a surgical patient's VO2 while under balanced anesthesia shows a practical application of the system.     

Study of on-line adaptive discriminant analysis for EEG-based brain computer interfaces

A study of different on-line adaptive classifiers, using various feature types is presented. Motor imagery brain computer interface (BCI) experiments were carried out with 18 naive able-bodied subjects. Experiments were done with three two-class, cue-based, electroencephalogram (EEG)-based systems. Two continuously adaptive classifiers were tested: adaptive quadratic and linear discriminant analysis. Three feature types were analyzed, adaptive autoregressive parameters, logarithmic band power estimates and the concatenation of both. Results show that all systems are stable and that the concatenation of features with continuously adaptive linear discriminant analysis classifier is the best choice of all. Also, a comparison of the latter with a discontinuously updated linear discriminant analysis, carried out in on-line experiments with six subjects, showed that on-line adaptation performed significantly better than a discontinuous update. Finally a static subject-specific baseline was also provided and used to compare performance measurements of both types of adaptation.     

Derived fuzzy knowledge model for estimating the depth of anesthesia

Reliable and noninvasive monitoring of the depth of anesthesia (DOA) is highly desirable. Based on adaptive network-based fuzzy inference system (ANFIS) modeling, a derived fuzzy knowledge model is proposed for quantitatively estimating the DOA and validate it by 30 experiments using 15 dogs undergoing anesthesia with three different anesthetic regimens (propofol, isoflurane, and halothane). By eliciting fuzzy if-then rules, the model provides a way to address the DOA estimation problem by using electroencephalogram-derived parameters. The parameters include two new measures (complexity and regularity) extracted by nonlinear quantitative analyses, as well as spectral entropy. The model demonstrates good performance in discriminating awake and asleep states for three common anesthetic regimens (accuracy 90.3 % for propofol, 92.7 % for isoflurane, and 89.1% for halothane), real-time feasibility, and generalization ability (accuracy 85.9% across the three regimens). The proposed fuzzy knowledge model is a promising candidate as an effective tool for continuous assessment of the DOA.     

Monitoring the Level of Anesthesia by Automatic Analysis of Spontaneous EEG Activity

Recent advances in the fields of automatic EEG analysis and pattern recognition provide a valuable new perspective for reconsidering the question of whether or not the level of anesthesia can be reliably estimated by analyzing spontaneous EEG activity. The feasibility of developing a computer-based EEG pattern recognition system capable of continuously estimating the level of anesthesia of patients during surgical operations is investigated in this paper. Anesthetists were asked to define five clinically significant levels of anesthesia for a commonly used anesthetic in terms of meaningful non-EEG criteria. The subsequent development of various EEG pattern recognition systems in an attempt to reliably estimate the levels of anesthesia as determined by the non-EEG criteria is described. All such systems employ Bayes decision rule under the assumption that pattern features are statistically independent. System performance is evaluated in terms of the estimated probability of misclassification error. Systems based on the recognition of spectral or frequency-domain EEG patterns are compared to those based on the recognition of time-domain EEG patterns.     

Feature-based detection of the K-complex wave in the humanelectroencephalogram using neural networks

The main difficulties in reliable automated detection of the K-complex wave in EEG are its close similarity to other waves and the lack of specific characterization criteria. We present a feature-based detection approach using neural networks that provides good agreement with visual K-complex recognition: a sensitivity of 90% is obtained with about 8% false positives. The respective contribution of the features and that of the neural network is demonstrated by comparing the results to those obtained with i) raw EEG data presented to neural networks, and ii) features presented to Fisher's linear discriminant.     

基于深度时空特征融合的多通道运动想象EEG解码方法

脑电(EEG)是一种在临床上广泛应用的脑信息记录形式,其反映了脑活动中神经细胞放电产生的电场变化情况。脑电广泛应用于脑-机接口(BCI)系统。然而,研究表明脑电信息空间分辨率较低,这种缺陷可以综合分析多通道电极的脑电数据来弥补。为了从多通道数据中高效地获取到与运动想象任务相关的辨识特征,该文提出一种针对多通道脑电信息的卷积神经网络(MC-CNN)解码方法,先对预先选取好的多通道数据预处理后送入2维卷积神经网络(CNN)进行时间-空间特征提取,然后利用自动编码(AE)器把这些特征映射为具有辨识度的特征子空间,最后指导识别网络进行分类识别。实验结果表明,该文所提多通道空间特征提取和构建方法在运动想象脑电任务识别性能和效率上都具有较大优势。

     

EEG Data Classification for Mental State Analysis Using Wavelet Packet Transform and Gaussian Process Classifier

Stress is one of the most common problems that is faced by a majority of the students. Long-term stress can lead to serious health problems, for example, depression, heart disease, anxiety, and sleep disorder. This paper proposes an efficient stress level detection framework to detect the stress in students using Electroencephalogram (EEG) signals. The framework classifies stress into three levels; low stress, medium stress and high stress. In this experiment, EEG data is collected from six subjects by placing two electrodes in the prefrontal region. During each trial, the subject solves arithmetic questions under some time pressure. The EEG data is collected while the subject solves the question. The collected data is pre-processed using a band-pass filter to remove artefacts and appropriate features are extracted through the wavelet packet transform and PyEEG module. ReliefF feature selection method is used to select the best features for classification. The selected feature set is classified into three categories using Gaussian Classification. The proposed framework effectively classifies the level of stress with an accuracy of 94%.

     

Pattern recognition techniques for the detection of epileptic transients in EEG

Epileptic transients (ET) in EEG, characterized by their amplitude and duration, are compared to background activity by means of cluster analysis in a two-dimensional amplitude-duration sample space. Decisions between ET and background are then carried out by assigning an optimum boundary between the clusters. By filtering the EEG signal using linear prediction followed by second differentiation, a useful degree of cluster separation is achieved. The probability of correct decisions is thus enhanced. Data for several patients are shown.     

BCI Competition 2003--Data set IV: an algorithm based on CSSD and FDA for classifying single-trial EEG

This paper presents an algorithm for classifying single-trial electroencephalogram (EEG) during the preparation of self-paced tapping. It combines common spatial subspace decomposition with Fisher discriminant analysis to extract features from multichannel EEG. Three features are obtained based on Bereitschaftspotential and event-related desynchronization. Finally, a perceptron neural network is trained as the classifier. This algorithm was applied to the data set (self-paced 1s) of "BCI Competition 2003" with a classification accuracy of 84% on the test set.     

Adaptive epileptic seizure prediction system

Current epileptic seizure "prediction" algorithms are generally based on the knowledge of seizure occurring time and analyze the electroencephalogram (EEG) recordings retrospectively. It is then obvious that, although these analyses provide evidence of brain activity changes prior to epileptic seizures, they cannot be applied to develop implantable devices for diagnostic and therapeutic purposes. In this paper, we describe an adaptive procedure to prospectively analyze continuous, long-term EEG recordings when only the occurring time of the first seizure is known. The algorithm is based on the convergence and divergence of short-term maximum Lyapunov exponents (STLmax) among critical electrode sites selected adaptively. A warning of an impending seizure is then issued. Global optimization techniques are applied for selecting the critical groups of electrode sites. The adaptive seizure prediction algorithm (ASPA) was tested in continuous 0.76 to 5.84 days intracranial EEG recordings from a group of five patients with refractory temporal lobe epilepsy. A fixed parameter setting applied to all cases predicted 82% of seizures with a false prediction rate of 0.16/h. Seizure warnings occurred an average of 71.7 min before ictal onset. Similar results were produced by dividing the available EEG recordings into half training and testing portions. Optimizing the parameters for individual patients improved sensitivity (84% overall) and reduced false prediction rate (0.12/h overall). These results indicate that ASPA can be applied to implantable devices for diagnostic and therapeutic purposes.     

Cardiorespiratory-based sleep staging in subjects with obstructive sleep apnea

A cardiorespiratory-based automatic sleep staging system for subjects with sleep-disordered breathing is described. A simplified three-state system is used: Wakefulness (W), rapid eye movement (REM) sleep (R), and non-REM sleep (S). The system scores the sleep stages in standard 30-s epochs. A number of features associated with the epoch RR-intervals, an inductance plethysmography estimate of rib cage respiratory effort, and an electrocardiogram-derived respiration (EDR) signal were investigated. A subject-specific quadratic discriminant classifier was trained, randomly choosing 20% of the subject's epochs (in appropriate proportions of W, S and R) as the training data. The remaining 80% of epochs were presented to the classifier for testing. An estimated classification accuracy of 79% (Cohen's kappa value of 0.56) was achieved. When a similar subject-independent classifier was trained, using epochs from all other subjects as the training data, a drop in classification accuracy to 67% (kappa = 0.32) was observed. The subjects were further broken in groups of low apnoea-hypopnea index (AHI) and high AHI and the experiments repeated. The subject-specific classifier performed better on subjects with low AHI than high AHI; the performance of the subject-independent classifier is not correlated with AHI. For comparison an electroencephalograms (EEGs)-based classifier was trained utilizing several standard EEG features. The subject-specific classifier yielded an accuracy of 87% (kappa = 0.75), and an accuracy of 84% (kappa = 0.68) was obtained for the subject-independent classifier, indicating that EEG features are quite robust across subjects. We conclude that the cardiorespiratory signals provide moderate sleep-staging accuracy, however, features exhibit significant subject dependence which presents potential limits to the use of these signals in a general subject-independent sleep staging system.     

Finger-based personal authentication: a comparison of feature-extraction methods based on principal component analysis, most discriminant features and regularised-direct linear discriminant analysis

In this study, feature-extraction methods based on principal component analysis, most discriminant features, and regularised-direct linear discriminant analysis (RD-LDA) are tested and compared in an experimental finger-based personal authentication system. The system is multimodal and based on features extracted from eight regions of the hand: four fingerprints (the prints of the finger tips) and four digitprints (the prints of the fingers between the first and third phalanges). All of the regions are extracted from one-shot grey-level images of the palmar surface of four fingers of the right hand. The identification and verification experiments were conducted on a database consisting of 1840 finger images (184 people). The experiments showed that the best results were obtained with the RD-LDA-based feature-extraction method 2 99.98% correct identification for 920 tests and an equal error rate of 0.01% for 64170 verification tests.     

An optimized index of human cardiovascular adaptation to simulatedweightlessness

Prolonged exposure to weightlessness is known to produce a variety of cardiovascular changes, some of which may influence the astronaut's performance during a mission. In order to find a reliable indicator of cardiovascular adaptation to weightlessness, we analyzed data from nine male subjects after a 24-hour period of normal activity and after a period of simulated weightlessness produced by two hours in a launch position followed by 20 hours of 6° head-down tilt plus pharmacologically induced diuresis (furosemide). Heart rate, arterial pressure, thoracic fluid index, and radial flow were analyzed. Autoregressive spectral estimation and decomposition were used to obtain the spectral components of each variable from the subjects in the supine position during pre- and post-simulated weightlessness. We found a significant decrease in heart rate power and an increase in thoracic fluid index power in the high frequency region (0.2-0.45 Hz) and significant increases in radial flow and arterial pressure powers in the low frequency region (<0.2 Hz) in response to simulated weightlessness. However, due to the variability among subjects, any single variable appeared limited as a dependable index of cardiovascular adaptation to weightlessness. The backward elimination algorithm was then used to select the best discriminatory features from these spectral components. Fisher's linear discriminant and Bayes' quadratic discriminant were used to combine the selected features to obtain an optimal index of adaptation to simulated weightlessness. Results showed that both techniques provided improved discriminant performance over any single variable and thus have the potential for use as an index to track adaptation and prescribe countermeasures to the effects of weightlessness     

Comparison of pattern recognition methods for computer-assistedclassification of spectra of heart sounds in patients with a porcinebioprosthetic valve implanted in the mitral position

The diagnostic performance of two pattern recognition methods (or classifiers) to detect valvular degeneration was evaluated in 48 patients with a porcine bioprosthetic heart valve inserted in the mitral position. Twenty patients had a normal porcine bioprosthetic valve and 28 patients had a degenerated bioprosthetic valve. One method was based on the Gaussian-Bayes model and the second on the "nearest neighbor" algorithm using three distance measurements. Eighteen diagnostic features were extracted from the sound spectrum of each patient and, for each method, a two-class supervised learning approach was used to determine the most discriminant diagnostic patterns composed of 6 features or less. The probability of error of the classifiers was estimated with the leave-one-out approach. The performance of each method to discriminate between normal and degenerated bioprosthetic valves was verified by clinical evaluation of the valves. The best performance in evaluation of the sound spectrum (98% correct classifications) was obtained with the Bayes classifier and two patterns of six features each. The percentage of false positive classifications of valve degeneration was 0% and the percentage of false negative classifications was 4%. Sensitivity for the detection of valve degeneration was 96%, specificity was 100%, positive predictive value was 100%, and negative predictive value was 95%. The best performance of the nearest neighbor method (94% correct classifications) was obtained by using the Mahalanobis distance and five patterns composed of three, four, five, or six diagnostic features. Using a pattern composed of only three features, the percentage of false positive classifications for degeneration was 10% and the percentage of false negative classifications was 4%.(ABSTRACT TRUNCATED AT 250 WORDS)