Automatic control of the administration of general anesthesia is considered useful for improving the safety of anesthesia and helping anesthesiologists with monitoring work. Basically, any control system should use multi-variables from various physiological systems, such as cardiovascular, respiratory, central nervous and neuromuscular, to control multi-agents such as anesthetic agents, drugs and fluids. The research currently performed is the beginning of this process, a system for the control of one agent, halothane, by one variable, mean arterial pressure. The control system is a discrete, digital type and is implemented using a small computer. The control algorithm is a combination of two control formulas. One is an optimal control to maintain the actual pressure at its reference level under possible minimum halothane administration, and the other is an adaptive offset adjustment formula to compensate for a time-varying and nonlinear aspects of a system. Results of testing the control system with dog experiments are presented.

抄録全体を表示

In this paper, a method of linear discriminant function which minimizes the error rate is introduced and a simple algorithm obtaining the linear coefficients which minimizes the number of misclassifications is proposed.
We applied this method to the Cephalo-Pelvic-Disproportion (C. P. D.) data composed of 2 groups-a group of 180 samples of natural delivery and a group of 60 samples of Caesarean operation. We compared the results with ones gained by other methods, such as Fisher's method, Walker-Duncan's multiple logistic method, Anderson-Bahadur's method and quadratic method. The number of misclassifications by our method was 12. That of Fisher's method was 18. That of multiple logistic method was 17. Those of Anderson-Bahadur and the quadratic methods were 16.

抄録全体を表示

The procedure for epicardial mapping consists of three stages : [1] recording of epicardial and reference signals, [2] extracting of the time of epicardial excitation and the reference time, and computing of the time difference (transmission time) between them, [3] displaying of epicardial map. At present, these procedures are executed by manual operation, and the results obtained in the stages [2] and [3] are influenced by subjectivity. To remove these difficulties, we have developed a minicomputer-based system for automatic processing of the stages [2] and [3].
The epicardial map is generally displayed by equi-time lines on the plane which expresses heart surface. In our system, two types of shape are used to express the heart surface. One is the expression conventionally used in manual operation, and, the other is the polar coordinate expression which we have proposed. The equi-time lines are drawn as contours of a two-dimensional mathematical function that represents epicardial excitation pattern. This function is determined by the least mean square method from the extracted transmission time at the stage [2]. The extracted transmission time is considered to be accompanied with some errors, so it is reasonable to apply the least mean square method to determine the function.
Collecting the epicardial potential waves from 65 points on the heart surface of a dog, we extracted the excitation transmission time by both the manual operation and the computer calculation. The transmission time difference at one point between the two methods was within 3 msec on 56 of the 65 points. Maximum difference was 8 msec at two points. In spite of the existence of these differences, the obtained epicardial maps by both methods were similar, and showed the normal excitation pattern.

抄録全体を表示

Analysing the mandibular movements is important in dentistry, as there are some problems in clinical application, this method has not been as popular as the EMG. The measuring equipment have so far included complex instruments to align the coordinates and have been difficult to operate. Furthermore, as is usual with biological information, it takes much time and effort to manage the measured data.
For this reason, a non-invasive system has been developed for this purpose with a transducer composed of LEDs, and cemented to the upper and the lower incisor teeth, and a photo diode which senses the position (x, y) of the input light spot. This measuring equipment determines the lower jaw displacements relative to the upper jaw by electronic means ; it is, therefore, not necessary to provide instrumentation to fix the head. To estimate the accuracy of measurement, the errors resulting from the non-linearity of the sensor, the variation of the distance between LEDs and photo diode, from rotation, inclination and other movements of the subject's head are discussed.
Our aim is the development of an examination system comprising a mini-computer, which analysis the masticatory system measured by our equipment and other devices for EMG and TMJ sound.
The computer system for the processing of biological information is composed of a 64 kw CPU, a 16-channel, 12-bit A/D-converter, disk storage, graphic display, and other peripherals. It has capability to aquire a great amount of measured data and to store them on the disk at the same time.
We have developed a software to process the measured data interactively and to display the movements as a function of time or in perspective using the 3 standard planes (frontal, sagittal, horizontal which could be partly magnified); velocity and acceleration of any component of themovements are also displayed. Furthermore, a section display that shows sectional diagrams of movements and a stereo display composed of a series of horizontal sections could be provided. Patient information such as name, age, sex, etc. and measured data relevant to the diagnosis will be selectively stored on a diskette (264 kB).
Our system could be applied clinically for the analysis of mandibular movements and should be useful in the diagnosis of masticatory function for its ability to manage measured data conveniently.

抄録全体を表示

Spatio-temporal filter theory is applied, on the basis of the describing function method, to anidealized nonlinear neuronal circuit of recurrently inhibited layer structure offering a model of synchronous neuronal activities in the brain.
The filter theory suggests that the system dynamics is fundamentally determined by the Fourier transform of the spatio-temporal connection in the system and that the increase in connection weight and feedback delay may cause synchronous oscillation.
The describing function for two sinusoids plus bias input reveals the competition among spatio-temporal dynamic modes such as synchronous oscillation, spatial waves, etc
Computer simulation as well as the describing function analysis shows that abrupt mode transition occurs between a spatial wave and synchronous oscillation with either internal parameter changes or appropriate external inputs.
These competitive relations among dynamic modes suggest possible connection with sudden mode transitions in EEG or in epilepsy.

抄録全体を表示