The configuration of a clinical laboratory system under the hospital-wide computer service network, including medical database system, is reported. The outline of data handling in the information network is shown, and evaluation of the contributions to daily clinical work and medical research by three years' investigation on database retrieval is made.
The biochemical, hematological, serological and urinoanalytic data are entered through automatic analyzers and VDT terminals. Those data are sent to medical database (MUMPS data file) and accumulated for more than three years. The daily laboratory reporting, the retrieval of data in the various forms, including trend graphs and statistical analysis for medical research, are all executed through the terminals provided at various sites of the hospital. Owing to the high efficiency of data compression of MUMPS, the system is able to keep the laboratory data on the disk storage for more than five years.
The capacity of laboratory increased about 60% as a result of the system installation. The retrieval of laboratory data up to 38000 times per year (1983) have suggested the high utility of this system, and it has been proved that the system on the network environment has been more efficient than the stand-alone type.

View full abstract

A practical, small and low-cost diagnostic information management system has been developed for a comparative study of various medical imaging procedures, including ordinary radiography, X-ray computed tomography, emission computed tomography, and so forth. The purpose of the system is to effectively manage the original image data files and diagnostic descriptions during the various imaging procedures. A diagnostic description of each imaging procedure for each patient is made on a hand-sort punched-card with line-drawings and ordinary medical terminology and then coded and computerized using Index for Roentgen Diagnoses (American College of Radiology). A database management software (DB Master) on a personal computer (Apple II) is used for searching for patients' records on hand-sort punched-cards and finally original medical images. Discussed are realistic use of medical images and an effective form of diagnostic descriptions.

View full abstract

This paper describes a new evaluation technique of impedance pneumographic field using the finite element method. The field is calculated on the basis of the inhomogeneous torso model comprising lungs, heart and thoracic layer. In addition, regional changes of the resistivity of lungs are respected.
It was found that the localized impedance change calculated at each lead and its distribution closely correlated with the regional change of lung's resistivity. It was also shown that the lower resistivity of thoracic layer resulted in a lower detection sensitivity.

View full abstract

Somatically Evoked Field (SEF) generated by a peripheral nerve stimulation has been studied using a SQUID with a second-derivative gradiometer in a room without magnetic shield. A three-phase signal is observed at the posterior region of the neck when an ulnar nerve is stimulated by electric current pulses. The signal appears during 30-150ms after the stimulation with a peak-to-peak amplitude of 1-1.5pT. The polarity of the three-phase signal reverses on the right and left sides of the midline of the neck, reaching the maximum at a distance of about 3cm from the midline. The amplitude saturates when the stimulation current exceeds about 8mA.
The SEF signal is also observed at the medial region of the suprascapula. A two-phase signal appears, in addition to the three-phase signal, depending on the measuring point.
As the signal sources for the SEF at the neck minute current elements, i. e., current dipoles, with a magnitude of 20-70nA·m are estimated at about 3.5cm below the surface of the neck in parallel with the midline.

View full abstract

There have been several studies undertaken of A/K (above-knee) prostheses capable of automatically adjusting the damping moment around the knee joint whereby an amputee may walk with his voluntary walking period. To control these A/K prostheses, the next step's walking period should be predicted in advance by estimating the amputee's intention during walking. EMG (electromyogram) was then introduced as a means of transmission of an amputee's intention. Taking into consideration the fact that limbs are activated by cooperative work of a flexor and an extensor, the EMG should be picked up from two location—one from a flexor and the other from an extensor. Hence the approach at predicting the next step's walking period by 2-channel EMG data has been performed.
The digital filtering algorithm for eliminating the low frequency noise on EMG was designed.
The recursive prediction algorithm by 2-channel EMG data was designed.
The experiment to predict the next step's walking period was performed by EMG picked up from various locations on normal person's thigh during level walking. From the results, it was found that the proposed prediction algorithm predicted the next step's walking period precisely when EMGs were picked up from the combination of M. tensor fasciae latae and M. gracilis.

View full abstract

Nerve conduction velocity (NCV) measurements have been widely used to assess the electrophysiological properties of peripheral nerves and to detect neuropathies at a subclinical stage. But conventional NCVs are usually expressed as the NCV for the fastest conducting fiber and the current standard methods do not supply information on the slower conducting fibers, nor detect information in individual fiber groups.
A new analytical method is presented for estimating the distribution of conduction velocity (DCV), based upon the spectrum analysis of the waveforms of two compound action potentials (CAPs) recorded by surface electrodes from a nerve bundle.
If the spectrums of the two CAPs recorded at two different sites in response to supramaximal electrical stimulation at the distances l₁ and l₂ are given as G_l1(ω) and G_l2(ω) respectively, the spectrum representation of the latency distribution P_l2(ω) for the propagation distance l₂ is expressed as follows:
P_l2(l₁ω/l₂)={G_l1(ω)/G_l2(ω)}P_l2(ω), where ω is an angular frequency.
For this formula, the algorithm which computes P_l2(ω) successively without using the iterative calculation methods is given.
Our estimation method is based upon the principle that the CAPs are recorded monopolarly to estimate the DCV, but in practical use, it is almost impossible to obtain appropriate CAP waveforms by the monopolar recording method, because of stimulation and muscle artefacts.
In order to evaluate the efficacy of bipolarly recorded two CAP waveforms for this computation algorithm, the CAP waveforms reconstructed by simulation techniques have been applied. As a result, it was found that the difference between the recording methods was reflected on the waveform of single fiber action potential but not on the latency distribution. The distance between bipolarly recorded electrodes did not affect the reproducibility of the latency distribution estimations.
It seems that this new method is non-invasive and could be used for evaluation of peripheral neuropathies.

View full abstract

Skin sensation has been employed to transmit pictorial information for the visually impaired, because it is spatially extended over the entire human body surface. Particular properties of the skin sensation have been revealed partially so far. In this study, pattern recognition through vision and skin sensation elicited by electrical stimulation has been investigated by means of raster scanning pattern presentation. Recognition rates for the changes of scanning speed were measured to compare pattern perception of the skin sensation with that of the vision.
In vision, higher recognition rates were obtained for fast scanning, which, however, decreased as scanning speed became lower. On the other hand, recognition rates of the skin sensation were much lower for fast scanning, and approximately equal to those of vision for slow scanning speed. Slow scanning improved spatial pattern perception through the skin sensation by electrical stimulation, but it is difficult to improve recognition rates for dot matrix alphanumerics through the skin sensation to over 80%.

View full abstract

In order to develop a new artificial neuron, a neuron model which is based on the many new findings about the neuron, especially on the neurochemical structure in a synapse, has been designed. The neuron model consists of four kinds of parameter; the absolute refractory period, the threshold function, the post-synaptic potential (PSP) waveform and the membrane potential function. We have examined effects of these parameters on the correlation between input and output of the neuron by the use of histograms and Markov property of impulse intervals.
Both histograms and Markov property of the output impulse intervals depended largely on the parameters related to the PSP waveform and the membrane potential function. It can be considered that a neuron has not only a threshold logic function but also the function of modulating input impulse intervals. This suggests that impulse interval processing function and the threshold logic function are highly important for the understanding of neuron's functions.

View full abstract