In this paper, we propose a new ultrasonic computerized tomographic technique which uses plane wave as insonifying wave. Spatial resolution of this ultrasonic CT image is independent of direction and depth since this ultrasonic CT uses plane wave insonifying. The technique uses a conventional linear array transducer. Therefore, the technique can transfer from the conventional pulse-echo system easily. Furthermore, it is feasible to reconstruct a high resolution ultrasonic image by one-dimensional deconvolution filter.
This ultrasonic CT is investigated by simulation. Point scatterers and circle model images are reconstructed in the simulation. These results show that this ultrasonic CT can recognize two point scatterers at a distance of 0.1875mm in spite of low SNR (10dB).

抄録全体を表示

The aim of this study was to provide adequate home health monitoring in the home by using fully automated biosignal measurement to support daily health care and to improve quality of life. We attempted to develop a home health monitoring system that did not provide any restrictions during sleep. bathing or elimination. A set of room containing a bedroom. bathroom. and toilet room was constructed. The system consisted of monitoring devices and a computer terminal for collecting data. The data were automatically collected from the monitoring devices placed at the bed, bath, and toilet and were transferred to a data terminal in the bedroom and stored for further analysis. The data acquisition system is now being used for seven days and data are collected without any trouble.

抄録全体を表示

We have fabricated a 64 embedded microelectrode-array substrate using semiconductor technology to investigate the biological signal processing in brain by using cultured neural networks of fetal rat neocortex in vitro.
We analyzed temporal and spatial neural networks patterns cultured on electrode-array substrate and attempted to examine the network structure constituted by neurons and the propagating patterns of electrical activity induced by the electric stimulus. In the experiments, each microelectrode size was 30μm squared and 150μm spaced. For stimulation, one of the electrodes was selected and current pulses were applied through an isolated circuit. After the network was cultured in about 50 days, responses of neurons to electric stimulus were monitored extracellularly through 64-channel electrode array.
Data recorded at each electrode consist of several spike trains generated by different cells. Therefore, these trains were separated by using wavelet transform and template matching for each electrode. We referred the temporal patterns of generated spikes for each electrode to as “spike sequences”. Next, we compared with the spike sequences among multichannel data and visualized the cultured neural networks structure by identifying the directions of propagations and cell connections.

抄録全体を表示

For quantification of the pulse diagnosis, it is necessary to standardize the displacement of the tissue around the radial arteries. In this paper, soft tissues around the radial artery of the distal forearm of the three healthy adults were visualized with the use of ultrasonographic imaging. It was found that the changes in the radial artery diameter and in the dermis thickness could be measured under the application of the hold down pressure caused by the interaction of pulse diagnosis. Further, the spatial distribution of these changes can be visualized during pulse diagnosis by means of ultrasonographic color doppler images of a distal forearm. Through this study, the effect of the hold down pressure on the radial artery was visualized and its relationship to the pulse diagnosis was assessed.

抄録全体を表示

Alpha activities as a basic component of EEG (electroencephalogram) are mainly observed with eye-closed and reported state, and indicates rhythmic and diffused pattern on the scalp. Therefore analysis of the relation among many sequences of alpha activities measured at different positions on the scalp is expected to be useful not only for diagnosing psychiatric disease, but also for investigating mechanism of brain information processing by means of causality analysis, that is, macroscopic estimation of flow pattern within brain. Although directed coherence analysis has been proposed as a method for estimating the direction and magnitude of information flow between two sequences, superposition of results for each pair of sequences can not represent true relation among the whole sequences.
In this paper, we proposed the multidimensional directed coherence analysis by modifying two-channel formula in order to apply it to the analysis of multi-channel sequence of alpha activities. Results of simulation revealed that multidimensional directed coherence can indicate more quantitatively the relation among the multi-channel sequences compare with conventional two-channel formula Moreover, the proposed method was applied to the analysis of EEG data of normal volunteer and patient. Results show that the method can provide a useful diagnostic information by assessment of the signal flow patter within brain.

抄録全体を表示

In the case of hypertension patients, monitoring of change in blood pressure and the strain level of the sympathetic nervous system is essentially required. In this study, a method to estimate the systolic and diastolic blood pressure from time-domain parameters of the heart rate variability is proposed. Using this method, it may be possible to monitor blood pressure continuously without posture restriction for a prolonged period of time.
Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were estimated from time series of RR intervals during change in posture and compared with those monitored with a portable digital sphygmomanometer. From analysis of data collected on normal subject aged 21-75 years, it was found that DBP and SBP were able to estimate from the heart rate, the parameters evaluating the autonomic function or the autonomic regularity and the age.

抄録全体を表示

In recent years, MRI (Magnetic Resonance Imaging) systems have been actually requested to diagnose the moving parts of a human body such as the heart or the blood in angiography, where images should be taken in milliseconds instead of minutes. The gradient power amplifiers which are small but vital components for advanced MRI require higher output power capability as well as faster rise/fall dynamic response characteristics under a variety of specified current reference signals.
This paper presents a novel switch-mode gradient power amplifier using IGBTs which is connected in parallel with the conventional 4-switch full-bridge power conversion circuits at their inputs/outputs in order to realize further high-power density. To satisfy the design specifications which require minimized the ripple and improved the rise/fall dynamic response characteristics of the current in the gradient coil, a unique digital control scheme on the basis of optimal type-1 servo system is proposed and described in detail. The above effectiveness is discussed and evaluated through the computer-aided analysis. It is expected that the proposed techniques will highly enlarge the diagnostic target and improve the image quality of MRI.

抄録全体を表示

The method of a Magnetic Stimulation (M.S.) have been widely used in neurophysiological research and clinical diagnosis. This method is very usefull to stimulate the human motor area and the nerve system without pain. In this method, however, a large magnetic field was applied to the human body and peripheral nerve systems, so that the inspection of the safety aspects of this high power magnetic field is one of the very important problems.
In this report, an experimental system which we have developed to observe the effect of magnetic fields on the rat's behavior is described together with some meaningful experimental results.

抄録全体を表示

We have previously developed an Alternating Noise Canceling (ANC) method that can reduce normal mode noise and equi-status noise, but parameter design and working principle are not sufficiently clear. In this paper, we have clarified the method of parameter design and working principle of ANC. The parameters of ANC method in application for instrumentation are switching frequency and cutoff frequency of low pass filter. Then these parameters are determined using a maximum frequency of transmitting signals and a maximum frequency of noises in transmission line. The working principle can be clear by interpreting ANC method as an amplitude modulation by a square-carrier wave and by applying frequency-domain analysis for modulated waves.

抄録全体を表示

A learning system is proposed to explain the adaptive function of an eye movement consisting of compensatory and optokinetic reflex, and pursuit movements based on the brain anatomy and physiology. Thereby, the learning system is synthesized as an artificial neural network based on the structure and function of the biological neural network of flocculus. The role of neural paths into flocculus from stretch receptors of ocular muscles are discussed in detail from the viewpoint of system control engineering. The mathematical learning process is also shown taking into account the adaptive mechanism and the anatomical structure of vestibular nuclei. The experimental results through simulation confirm the validity of the hypothesis and the appropriateness of the inference process in connection with the proposed mathematical model.

抄録全体を表示

This paper presents the measurement of three-dimensional shapes of vocal tract and dental crown using magnetic resonance imaging (MRI). 3D MR images of the vocal tract of 3 adult males were obtained in the steady state productions of Japanese fricatives /s/, /sh/ and /shu/ using a dental crown plate contain a contrast medium for MR imaging. MR images and the voice output uttered by a supine subject in the MR system were obtained simultaneously. The measurement time was 186 second for 41 axial sections from the larynx to the nasal cavity at a 4mm interval, and was 186 second for 32 coronal sections from the tip of the nose to the atlas at a 4mm interval. Frequency characteristics of the voice output /shu/ showed results specificity in each subjects. 3D shapes and area function of vocal tract during the productions of Japanese fricatives /s/, /sh/, and /shu/ were obtained from arbitrary sections computed by gray level interpolation. Several narrow points that were considered articuration points were detected in vocal tract from these results of measurements.

抄録全体を表示

Recently, the crystal oscillator as a source of frequency standard have required miniaturization, low power consumption and high frequency stability because of the rapid development of the mobile communications represented by the cellular phones. It is necessary for PLL and so on to be composed of a VCXO (Voltage Controlled Crystal Oscillator). The VCXO should be implemented on CMOS-IC. Traditional VCXO is controlled by capacitance variation of a varactor diode. But it is difficult to realize the varactor diode on IC chip.
In our previous study, we showed that the transistor VCXO using the MOSFET's Miller capacitance had a wide frequency variation. But the Miller capacitance decreased with large amplitude. So, a wide frequency variation could not be realized the CMOS-VCXO using the Miller capacitance.
In this paper, first the variable capacitance circuit is realized in order to construct by the wide variable range CMOS-VCXO for IC. The wide variable capacitance circuit is composed of the MOSFET as the voltage controlled resistance. Next, the CMOS-VCXO is constructed by the variable capacitance circuit and the CMOS crystal oscillator. As a result, we will show that the CMOS-VCXO has wide frequency variation of about 400[ppm].

抄録全体を表示

In this paper, we propose a high-speed RSA encryption on a bit sliced system, using a number of the same multipliers and dividers easily expandable to any accuracy. Preparing the hardware for the speeding up in RSA encryption needs extremely a large integrated chip because of the necessary residue arithmetic for numbers more than 1024 bits long, and will bring up the cost.
The multiplier and divider developed by authors can perform these data processing to any accuracy. And it can be implemented on a single chip with 12, 000-gate for the circuit with 64-bit for multiplier, and 128-bit for divider. It can cope with the requirement for 1024-bit only by cascading 16 chips, within the time interval proportional to the accuracy. It can be implemented as the cryptosystem on each personal computers, and can be realized as RSA encryption desired for the information security. This bit sliced system will bring down the cost of the system implementation and we can expect much higher speed and safer cryptosystem as a tamper-resistant configuration because it is made of hardware for the most part.

抄録全体を表示

The ground area of print pattern is important for the reduction of EMI. But until now the ground area of the print pattern was not quantitativery examined. In this report we present three parameters relating to ground/power-source patterns which are important for the EMI reduction. One of them is a “BETA” ratio which is defined as the ground pattern ratio to the print board size. Our experiment shows that these ratios are greatly affect to the radiation noise.
So we developed the CAD, which is useful for print pattern design, to caluculate these parameters.

抄録全体を表示

Recently, for creating realistic computer graphics (CG), it has been increasingly requested to obtain 3DCG models (3D shapes and textures) from photographs and movies. We have proposed a system construction method which can extract 3DCG models by user's simple operation from a series of image taken with normal cameras and videos.
The method is consisted of two phases-to extract partial models and to combine those models for creating a complete model. On the first phase, the operator input some feature (points, edges, planes) to each image. On the second phase, the operator just select several pairs of feature points of the adjacent partial models. Additionally, this method has the function to correct the shapes of each model by changing only two parameters.
This paper represents the principles, procedure and operational examples of the proposed method. It also shows an experimental result of model extraction accuracy and discusses the characteristics of the proposed method.

抄録全体を表示

This paper presents high-speed and low power dissipation architecture for higher-order FIR filters with very small latency. To minimize power dissipation and latency, the distributed arithmetic, of which processing time depends on only word length, is applied to this architecture. The basic structure method of function Φ using ROMs has been proposed so far. However this method for function Φ requires very high power dissipation and wide area in VLSI chip, because of using ROM of much larger capacity for higher filter order. So, we propose a new implementation approach of the function Φ using logic gates. We also reveal a new method of lower power dissipation, by making use of a perfect linear phase characteristic of FIR filters. As a result, by using 0.8μm CMOS technology we show that a 120-tap FIR digital filter can be realized with high sampling rate of about 2.74MHz and the small latency of 572ns. These values are almost independent of filter order. Moreover this processor is implemented with power of 1.61W, which can decrease power dissipation to about 13.9% as compared with conventional implementation using multipliers.

抄録全体を表示

This paper proposes the learning methods to control the frequency of an oscillatory neural network. The learning rules are applied to the neural oscillator that comprises two excitatory neurons, in which only one neuron has a positive feedback weight. Since it is assumed that only the feedback parameter has plasticity, the proposed learning rule can be realized with a high simplicity. By defining the phase of the neural oscillation, a mathematical model is conceived so as to conjure up of the blurred vision of phase trajectories in the system. Successful examples of the frequency learning of the sinusoidal function is shown by the computer simulation. With the proposed learning methods, the frequency of the oscillatory neural network can be adjusted to that of any desired value.

抄録全体を表示

For the purpose of automatic and effective neural network structure design, a designing method is proposed based on Genetic Algorithms (GA). Neural Network structures are encoded on chromosomes in such a way that a number of different but related network structures result by changing a parameter in the decoding process. The GA finds an optimal chromosome that provides good network structures for a family of data sets. When a change takes place in the environment around the neural network, by changing the parameter, we can obtain a new neural network suitable for the new environment without re-running the GA again.
The proposed method is described by using, as an example, a problem of finding good neural network structures for data sets with different noise magnitudes. To obtain an accurate but non-overfitted neural network for the noisy data set, we introduce a relevant fitness function, a method for noise magnitude estimation, and a systematic way to determine the control parameter value of the decoding process. By incorporating these techniques in the network structure designing method mentioned above, we obtain a neural network which has a good generalization ability for each of data sets with different noise magnitudes.

抄録全体を表示

Usually, analysis of a conventional symmetrical 3-phase feed-back control system can be done using a single-phase conventional transfer function. However, analysis of 3-phase system having rotating coordinate transformations is difficult by the conventional transfer function. Because, the conventional transfer function cannot deal with the control system having the rotating coordinate transformation inserted in the feed-back loop. Therefore, in this paper, the authors propose a novel extended transfer function with the rotating coordinate transformations and analyze the system having a rotating coordinate transformations.

抄録全体を表示

This paper focuses the maintainability of the workflow systems. In order to make the workflow systems follow various changes of the business environment, it is important to exactly separate the stable part and the changeable part in the system. A new concept-“business process design pattern”is introduced. In this concept, the business process is considered as commitment network among actors related to the target work, and the basic tasks and their orders executed by the main two actors compose the stable process structure-process pattern. On the basis of input/output data of the basic tasks, the necessary and enough information structure for the target work is extracted-information pattern.
The design method of the business process using these design patterns is proposed, by which the individuality of process and application is excluded, and the high maintainability of the workflow system is gained. By the case study of the order management work, the effect of this method is shown.

抄録全体を表示

This work investigates a new approach to the tracking of multiple moving objects. The method comprises of two level procedures. At the first level, it tracks each object by comparing its shapes between successive frames using Hausdorff distance. At the second level, it uses context information to reduce the complexity of the problem and break it into several more manageable ones. It means that the condition about each object (in this case: correspondence, state, etc.) will be evaluated at each frame and the proper tracking procedure will be initiated based on its evaluation at each object basis. The main contribution of our work is the development of searching and updating procedures and also objects management procedure which makes the system capable to track objects under various, real condition. Experiments conducted on real images demonstrate that the approach is robust against overlapping and various objects' condition.

抄録全体を表示

A prediction method of plant status using neural network has been proposed for operation support systems in plants. The change of time series data from plant is complex because the data contain many frequency components. It is difficult to improve prediction performance of neural network by using the time series data directly. The proposed method resolves the data into some components based on their change frequencies and a prediction model using a neural network is generated for each components individually.
To evaluate the proposed method, two kinds of time series data are trained and predicted. One is the data containing several periodic components and a non-periodic component generated by an equation. The other is the simulation data of a fossile power plant under an abnormal condition. As a result, time averaged prediction error was reduced to 1/4-1/3 compared with a method using time series data directly. Thus, the proposed method is effective to predict plant status precisely.

抄録全体を表示

The condition of surface roughness affects the quality of goods, because the light-distribution reflected from the surface depends on the roughness. The relationship between the reflected light-distribution and the roughness, however, does not be made clear. In this study, it is investigated that the correlation between the reflected light-distribution and the parameters representing roughness defined by JISB0601. At first, a simulation is practiced. A roughly finished surface model is designed and the reflected light-distribution is calculated by ray tracing. The relationship between the reflected light-distribution and the parameters representing roughness is examined by the use of the simulated results. As the results, it is found out that the reflected light-distribution is not much correlated with each roughness parameter alone but has a high correlation with the ratio of Ra (arithmetical average roughness) to Sm (mean interval of surface unevenness). For the validity of the simulation, the reflected light-distribution and the parameters representing the surface roughness are really measured on roughly finished plastics. From the experimental results, the same relationship as the simulation is obtained. In addition, the reason why the reflected light-distribution is described as Ra/Sm is mentioned by the use of a surface model.

抄録全体を表示

This article explores a proposed method for identifying and classifying the color uniformity defect on the electronic displays. A neural network learning approach utilizing the backpropagation learning algorithm is employed to search the dissimilarity of the color display condition. The color uniformity defect image perception among several conditions of some observations by human eyesight is used to supervise the training data of the neural networks. A supervisor of the training data is obtained by human expert eyes evaluate the white uniformity grade and compare with the standard grade, then labeling the perception grade into some certainty values. The trained network is used to identify and classify the grades level of the color uniformity defect on the electronic displays. For the experimental purpose, a simulation program is developed to generate the color uniformity defect on the monitor display. We make a comparison with a classical regression analysis method for validation of the proposed method.

抄録全体を表示

A doubly coprime factorization plays an important role in the design of proper stabilizing controller for given LTI systems. In this paper, using polynomial matrix approach, a reduced-order doubly coprime factorization is considered. Each element of the proposed factorization is proper transfer function matrix. Sufficient conditions for controller reduction are also given.

抄録全体を表示

During the past decade, several specific hardware systems for fast fuzzy inference have been developed. However, most of them are dedicated to be executed sequentially, rule by rule, although they have parallel processing manners in operations of AND/OR computations. In this paper, we propose a parallel inference method of processing several fuzzy rules simultaneously by using T800 Transputer system, and analyze the worst case upper bound for internode message delays. The proposed control system has a relatively higher speedup, compared with the results of experiments in a single processor. This parallel inferencing fuzzy system can be used in a system requiring a rapid inference time in real-time system and/or one that has many inference variables in the condition part and the consequent part. Also, this system will be applied to build powerful architectures for control applications like robot control with time-critical sensor integration.

抄録全体を表示

Several control models are proposed and developed to distribute water in proper pressures and flows. The optimum control model of water distribution is the best one of them from the viewpoint of the cooperative operations and the controllability of global area. The optimum control method with pipe network analysis is a kind of known method having those advantages. But it doesn't work well in practical systems because it has no means of estimating the spatial variations of water demand in real time, which is the load upset of the systems. The proposed method resolves this problem by the optimizing method using monitored pressures and flows in the pipe network.
First, the problem of estimating the spatial variations of water demand is formulated. Then shown is the result of experiments in a practical system based on this formulation. Finally, the validity of this estimation is mentioned by being compared with the geographical data about the land use.

抄録全体を表示

In this paper, we present parallelizing techniques for the two fuzzy inference methods (Mamdani and Koczy methods), and propose an effective parallel fuzzy architecture, which is suitable for fuzzy information processing. The proposed parallel fuzzy architecture is composed of FPEs, CP, memory modules, interconnection network and MIN circuit. This achieves a relatively higher performance and is a generalized cascadable architecture independent of input variables, output variables and rule base. In this, each FPE_i processes only the operations of the i-th antecedent and the i-th consequent. Then, all of the antecedents in the condition part are executed in parallel, same as in the consequent part. So processors can be fully utilized and it is easy to configure the design with any number of antecedents, consequents and rules. This parallel fuzzy inference architecture can be used in a system requiring a rapid inference time in real-time system and/or a large expert system that has many inference variables in the condition part and the consequent part.

抄録全体を表示

A problem on an optimization of fuzzy reasoning presents two aspects, a numerical optimization problem and an order optimization problem. The genetic algorithm (GA) is able to be applied to both these problems. In many research works, though the configuration of the fuzzy reasoning is optimized by using the GA, shapes of the membership functions are optimized by using another optimizing algorithm, such as a steepest descent method. In the background of this method, since a mutation operator of the GA always globally searches in a solution space, the GA optimizes slowly. In the case of applying the fuzzy reasoning to a field of control, the global optimum is not always required. There are many cases that some local optima give appropriate performance. To obtain such local optima, the authors have proposed a variable bit-selection probability (VBSP) in process of the mutation operator. The VBSP allows the GA to search around a good solution candidate. In this method the bit-selection probability (BSP) is partially defined for each bit position and varied according to the fitness value. However, this method requires designer's experiences and preliminary implementation in a phase of designing in some cases. To improve this difficulty, we propose to linearly vary the VBSP according to the number of the generations in this paper. By means of this new method, the GA with the VBSP becomes a general purpose algorithm that the designer's experiences and the preliminary implementation are not required.

抄録全体を表示

Nowadays various kinds of information is flowing in the Internet Especially huge amount of information is available on the WWW (World Wide Web) and it contains useful technical information at the high rate, which we can hope to be increasing. But WWW has two problems caused by the flood of information, the problem of the information searching (difficulty of getting information on WWW) and the problem of the information filing (difficulty of arranging the information got on the Internet). They prevent the efficient utilization of information on WWW. We have investigated the measures against them and developed Technical Information Filing System (TIFS). TIFS is the practical environment for utilization of information on WWW which supports gathering, registration and retrieval of the information.

抄録全体を表示