In using the balanced non-cephalic reference electrode (BNE) system, ECG artifacts mixed in recorded EEGs were not easy to reject in the patients with left axis deviation of the heart.
In this paper, we have presented the effects for ECG rejection due to the changes of the balanced resistance value, because the ECG rejection was most dependent on both the value of electrode contact impedance and the balanced resistance value, and the impedance converter to reduce the high electrode contact impedance was also most available for ECG rejection in the BNE system.
The difficulty of ECG rejection in the patient with left axis deviation was proved from measurements of both the cross-correlation and linear regression coefficient between the ECGs at the electrode on head and those at the reference electrodes of BNE system. But when the two reference electrodes of BNE system were placed at both sides of the right anterolateral side and the left posterolateral side on the neck, it was possible to reject the ECG.
Conclusion : ECG artifacts can be rejected much more by the use of both the BNE system with new position of reference electrodes and the impedance converter.

View full abstract

In order to get a better model of the substitutional sensory device for the blind, some characteristics of the tactile sense of a fingertip have been investigated based on psychophysical experiments with respects to the maximum information transmission, the 2-point threshold and the recognition of Braille.
Raised dot patterns, which had the 0.7 mm stimulation area, various heights (0 to 0.5 mm) and various arrangements, were displayed on a fingertip surface by sliding the fingertip or the pattern, and experimental data were obtained as a function of the sliding speed.
From experimental results, 1) the maximum transmission information by a dot was found to be about 1.8 bits/stimulus at 4 cm/sec sliding speed. However, it decreased remarkably when the speed was beyond 20 cm/sec. 2) The 2-point threshold showed about 1.5 mm at 4 cm/sec, but it was about 2 mm at the sliding speed between 6 and 20 cm/sec. 3) The identification rate of 7 Brailles was about 80% at 3 cm/sec speed, though it also decreased at the speed beyond 4 cm/sec.
By comparing the perception characteristics of vibratory patterns with those of raised dot ones, physiological mechanism of the raised dot perception has been discussed from a view point of the characteristics of the mechano-receptors such as pacinian and non-pacinian.

View full abstract

This paper presents a study on the categorization of the grade of pneumoconiosis. Usually, texture analysis techniques are used for measurement of profusion of opacities in radiographs. In this paper, features of texture patterns were extracted from the analysis of statistical characteristics of one-dimensional density sequences obtained by scanning X-ray pictures. Three scanning directions were considered and extracted features from each direction were put together to evaluate statistical characteristics of two-dimensional patterns. For these purposes, the linear prediction method was applied.
The classification system based on the linear prediction method has an advantage of not being required to exclude rib space from the processing area which has usually a bad influence upon the analysis of profusion of opacities.
A four-category classification was performed for both the zonal samples and the overall films with correct classification rates of 73.8% and 78.6% respectively. Opacity type classification was also performed, and 96.3% of ILO standard films for rounded opacities were correctly classified. These experimental results show that the automatic classification system proposed in this paper is very effective.

View full abstract

Thoracic impedance changes, measured by the electrical impedance method, are composed of the impedance change associated with ventilation, ΔZ (resp) and the impedance change associated with cardiac pulsation, ΔZ (pulse). However, it is very difficult to measure one of them selectively. Therefore, it is indispensable to discriminate betweenΔZ (resp) andΔZ (pulse) and extract one which is necessary in accordance with the purpose.
This paper describes a method of eliminatingΔZ (pulse).ΔZ (pulse) superposed onΔZ (resp). causes errors in the ventilatory volume measurement by the electrical impedance method. Firstly, examination is carried out inΔZpower spectrum which changes with clinical course for patients. The analysed results are summarized as follows :
1.ΔZ (resp) is composed mainly of the fundamental frequency power, because the second harmonics power is about 18dB less than the fundamental frequency power.
2. The fundamental frequency power ofΔZ (pulse) is about 14 dB less than that ofΔZ (resp).
3. The fundamental frequency ofΔZ (resp) and that ofΔZ (pulse) show individual differences, and even for a patient they change with time. However, in most cases the ratio of the fundamental frequency ofΔZ (resp) to that ofΔZ (pulse) is maintained at higher than three or four.
Based on these results, a filter for eliminatingΔZ (pulse) has newly been developed. This. is a low-pass filter whose resonance frequency is controlled by the heart rate, which ralates to. the fundamental frequency ofΔZ (pulse). Accordingly, this filter has flexibility of being able to follow the variation in biological condition, and is very practicable for clinical use.

View full abstract

A body surface potential mapping system equipped with a microprocessor has been developed.
The system enables high speed, real time and simultaneous observation of ECG data of 96 lead points in a character display mode. The machine has functions not only of making such various basic data processing as base line correction or linear interpolation for malfunctioned electrodes, but also of generating isopotential contour map, to record data onto floppy disk, to make hard copies of displayed patterns and characters, and so on. To increase the operation speed, all the assembler programs were put in a module. Thus, an operator can easily recognize the spatial motion pattern of the body surface potential map and in addition, find out malfunctioned channels by observing the pattern. Also a belt type electrode system allows quick and flexibly adaptable attachment.
The system has proved its practical usefulness under cooperation of cardiologists; over 100. patients with cardiac diseases have been recorded.
The system has a peculiar function to transmit the data to the other large computer center to make further high level data processing like inverse estimation of epicardial potentials.

View full abstract

This paper describes the automatic recognition method which classifies every short frame of the respiratory sounds. Discrimination of every short frame enables not only detection of adventitious sound, but also separation and classification of respiratory sounds which contain different sounds at different times.
Discrimination consists of two stages. First by means of the mean square of residual error, respiratory 'sounds are classified into two classes, continuous sounds and discontinuous ones. Secondly each class is subdivided by means of feature parameters based on AIC (Akaike Information Criterion) which are transformed from PARCOR (Partial Autocorrelation) coefficients. The standard category sets are defined by R. Murphy's medical training tape, and the effectiveness of this method is shown by the discrimination experiment.

View full abstract

This paper describes fnequency spectnal analysis of ECG f-wave and its application to the f-wave recognition in a computer diagnostic system (Nihon Kohden DCC-3300, CPU NOVA-01). Short interval signal of V₁ chest-lead in the standard 12 leads transmitted through telephone line was mainly analyzed by means of Fast Fourier Transform.
Fnequency spectrum of f-wave showed a peak modulus (f-peak) of more than 0.01±0.005mV in the range of 5-8Hz, the features of which were quite different from those without f-wave. It therefore follows that high accuracy recognition of f-wave would be possible by setting a frequency window in the range of 4.5-8.5 Hz and the threshold discriminating f-peak modulus. Particularly for the data composed of multiple f-wave segments, thershold of 0.006 mV has made possible the recognition of f-wave and classification of atrial fibrillation with more than 90% correctness.
Although recognition of f-wave within associated noise and other background signals is difficult, our proposed method should be useful for the recognition of f-wave accompanied with atrial fibrillation in a telephone-transmitted ECG diagnostic system.

View full abstract