As means of stimulating the neuron and measuring the action potential, intracellular methods with micro-pipette are ordinarily used in the field of neurophysiology. Having many merits, the intracellular methods are not suitable for a long-term experiment due to damage to membranes. Moreovere, for the case of intracellular method the multisite stimulation and measurement is difficult from the point of access to the neuron and its invasiveness. In this study, we investigate the extracellular neural stimulation system with arrayed microelectrodes for the purpose of studying the process of organizing neural networks and signal processing by neural cells. Using the microelectrode array of the size of cells formed on the substrate, the action potential of cultured neuron was recorded. The experimental results showed the feasibility of the extracellular stimulation electrodes. The patterns and position of electrodes, furthermore, for appropriate stimulation were examined by simulating the distribution of the current around the ax on model for the various cases. The results offerred some points for design of optimal electrodes.
It is necessary to make a model of blood concentration in order to give the optimal administration of medicines. We often assume a one-compartment linear model with absorption in case of oral administration. In this paper we propose two estimation methods of the unknown parameters of this model. They are completed with only one or two data of blood tests conducted at optimal times. This easy modeling will make the proposed methods more useful in medical treatments of diseases such as cancer.
The measurement of a hardness on the human body has been anticipated in a basic and clinical medicine, a skin science, and other various fields. Some indexes of hardness on the human body, which have been based on a dynamic measurement in vivo, have been reported, but there are few to lead to a practical application i. e. an evaluation of palpation. In order to realize a practical application, a hardness-meter for the human body has been developed. This measurement system uses a single sinusoidal vibration to simplify both its hardware and software. SH (Skin Hardness), which expresses mechanical mobility, is proposed as a new index of hardness. The measuring probe is designed as a handy type to measure almost all around the human body and the whole system is a portable type. The forced vibrating and the detection of acceleration are realized by a piezo-electric element of bi-morph type. Besides, the measuring error, the proper measuring conditions and the relations between SH and the other indexes are made clear.
The autonomic nervous system influences heart rate variability. The sympathetic and parasympathetic nervous system are usually the principal systems in short-term cardiovascular control. We analyzed the ΔR-R intervals of electrocardiogram using autoregressive model (AR model) to evaluate the autonomic nervous system function which influence the atrio-ventricular node. The parasympathetic nervous system is dominant at supine position and the sympathetic nervous system is dominant at standing position. Therefore we compared the first degree functional atrio ventricular block (AV-block) group with the control group at supine position and standing position. The control group (60 normal subjects) were compared by discriminant function test using AR-coefficients as variables. The control group were discriminated at standing position (p<0.01). Analysis of heart rate variability using AR model is useful to estimate the balance of the sympathetic and parasympathetic nervous systems.
This paper presents a rotation invariant neural pattern recognition system with an edge detection architecture. The system consists of an edge detection network and a trainable multilayered network. The edge detection network preprocesses an input pattern to produce the edge features, which are the activity pattern of orientation specificity cells (OSC), The OSC pattern is an input pattern to the multilayered network. This network can be invariant to rotation by any number of degrees. The network weights are completed through the back-propagation training. Finally, computer simulation for coin recognition shows the effectiveness of the system, and comparison of the system with the conventional one is described.
The present paper proposes a method to estimate the motion intended by a human operator from his EMG signals using the statistically structured neural network. EMG signals gradually appear at the beginning of the motion and begin to vanish at the end of the motion. Therefore they should be regarded as non-stationary signals and only weak assumptions are made about the probability density functions. In order to classify such non-stationary signals, the neural network presented here is statistically structured using a Gaussian mixture model which can approximate an unknown probability density function by a finite mixture of multivariate Gaussian component densities. The experimental results for non-stationary EMG signals show that the network can learn the unknown densities of the EMG signals and it can discriminate six motions of forearm and hand based on Bayesian rule from unlearned EMG patterns with the accuracy above 90%.
We have developed two-dimensional lock-in amplifier systems that can detect very small changes in the light intensity distributions of images. In optical microscope systems, the resolution is lower in the depth direction than in the focal plane, thus it is important to improve the depth resolution for high-resolution, three-dimensitnal imaging. In this paper, we present a new technique to improve the depth resolution by implementing the lock-in amplifier technology. In this technique, the derivatives of the image intensity in the depth direction are obtained using the lock-in amplifier system. Using these derivatives, the depth resolution can be improved by a factor of 3.5. We believe that this system has many important applications in the two-dimensional measurement of optical information.
In this paper, we proposed a tomographic image reconstruction system. Tomographic images are performed by a 35mm X-ray cinematogram obtained during a rotational motion of the cineangiographic apparatus. We can freely select the position of a tomographic plane with any position and angle. In the beginning, we took a phantom including 11 small metallic particles on different planes. The phantom images were taken over 180 degrees, taking 28 seconds at 30 frames/s using a 35mm X-ray cinematogram system without guaranteeing isocentric rotation of the X-ray source or image intensifiers. Assuming that the coordinates of these 11 particles could be determined, we obtained mathematical relationships between the 3-dimensional coordinates corresponding to the particles and the 2-dimensional coordinates of the particles' shadow on the cineangiographic images. Thus, the relationship between the coordinate of any point in bounded space and its location on a cineangiographic frame image could be determined. Next, a phantom of the cerebral artery blood vessels was taken using same methodology as described above. We reconstructed tomographic images by the back projection method using this mathematical relationships and obtained images were in sharp focus. We can freely and easily set tomographic planes, and reconstruct tomographic images with our system.
Although ultrasound images are clinically useful, they have artifacts since ultrasound does not travel in straight lines through the material due to refraction and diffraction effects within the complex structure of the body. Computed tomography considering wave refraction and diffraction effects is called “Diffraction Tomography”. In cases where the number of viewing angles is small, the quality of reconstructed diffracion tomogram using filtered back-propagation algorithm is better than that of X-ray CT using filtered back-projection algorithm. The iterative technique have been proposed for improving the quality of reconstructions from projections when the number of views are small or the angular range of views is limited. The technique consists of transformation repeatedly between image and spatial frequency domain and applying a priori object information at each iteration. We applied this technique to diffraction tomography using filtered back-propagation algorithm. In computer simulation results, it was shown to be possible to improve quality of reconstructions in a iteration.
The paper is concerned with improvement of the chirp radar-type microwave computed tomography developed for non-invasive thermometry of the human body. The hardware of the radar-type microwave CT is quite simple and the reconstruction algorithm which has been used for X-ray CT systems is applicable. This is because the microwave CT can discriminate the transmitted waves on the straight ray path between the transmitting antenna and the receiving antenna from multipath signals. We have already reported that the spatial resolution is about 1cm, but temperature resolution has never been directly measured because of its poor resolution of the FFT analyzer. The analyzer was replaced by the new model whose resolution and accuracy were enough for discriminating temperature change less than 1°C. The YIG-tuned bandpass filter was incorporated in the improved system along with the YIG-tuned oscillator which was used to generate the microwave chirp signals. This was very useful to improve the SN ratio of the tomograms and also useful to expand the dynamic range of the attenuation measurements. Throughout a series of the experiments using phantoms of a human body, temperature change by 2°C was successfully imaged. The temperature change by 1°C was not observable as the image for temperature instability of the system. But imaging of the temperature change which is less than 1°C may be feasible by shortening the time required for the measurement.
We have classified parenchymal echo patterns of cirrhotic liver into four types according to the size of hypoechoic nodular lesions. Neural network technique has been applied to the characterization of hepatic parenchymal diseases in ultrasonic B-scan textures. We employed a multilayer feedforward neural network utilizing the back-propagation algorithm. Coarse score of cirrhotic liver at autopsy correlates closely with histological findings. The coarse score is also bound up with parenchymal echo pattern of the liver. The experimental study suggests that the neural network approach is useful for objective evaluation of hepatic parenchymal echo pattern, though continuing work is needed to improve the classification accuracy.
This paper has proposed an algorithm by which the tissue can be extracted using texture information of ultrasonic echo signals which are obtained from the human body. The extraction is the inevitable process when any target tissue can be three-dimensionally imaged on the CRT. So far, the intensity thresholding method has been executed in which the great difference of echoes on the boundaries between fetus and amniotic fluid has been used for detection of the organs like a three-dimensional (3-D) display of the fetus. Therefore in case of tumors in the liver there is not always a definite difference between the tumor and normal tissue. So it is difficult to distinguish between them by the conventional thresholding method. The proposed method here, which is expanded from a conventional two-dimensional texture analysis to a 3-D one in order to advance the ability of extraction, makes use of the difference of texture information between them and has demonstrated the extraction results of the liver tumors.
Gap phenomena and discharge current Id are investigated experimentally under the conditions of krypton concentration GKr of 1 to 20% in (Kr/He) mixture gas and fluorine concentration GF2 of 0.1 to 1% in (F2/He) mixture gas to clarify influences of krypton and fluorine gas on discharge characteristics for a KrF excimer laser. When GKr is over 10% in (Kr/He) mixture gas, corresponding to the production of filamentation in glow discharge, Id is seen to be changed discontinuously in the second or third half cycle. Based on both experimental and numerical results, the discontinuity of the current is concluded to be a sign of the generation time of filamentation. In the case of 0.1_??_0.3% of GF2 in (F2/He) mixture gas, filamentation and arc with glow are observed. Id terminates in first half cycle owing to an electron attachment by fluorine gas, but begins to flow again after the termination or has a plateaus due to arc and filamentation. More than 0.4% of GF2, only arc is generated. Id becomes small and pulse width of the first half cycle is extended due to extreme contraction of discharge area. In the case of (Kr/F2/He) mixture gas, filamentation and arc with glow are observed and Id terminates in first half cycle. Fluorine gas is clarified to have a large effect on discharge and current characteristics in KrF laser mixture gas.
Estimating some mappings by neural networks, a part of mapping properties is often known beforehand. The back propagation type neural networks, however, do not utilize this kind of knowledge about the mapping. The present paper proposes to incorporate known nonlinear functions involved in the mapping into the back propagation type neural networks in order to utilize the knowledge about the mapping. As a result, the known and unknown parts of the mapping can be learned in preorganized and unorganized layers of the neural networks respectively. Then, the preorganized neural network is applied to inverse dynamics problems of robot manipulators. Experimental results show that the learning abilities such as convergence characteristics, generalization abilities and parameter identification can be improved compared to the conventional one by incorporating the motion equation of the manipulator into the preorganized layer.