In the external beam radiotherapy, it is required to optimize the number of beams, the beam angles and the strength of irradiation which deliver the desired dose to the tumour while limiting the dose absorbed in the normal tissue. Mathematical programming techniques are reported to be useful to solve this optimization problem with various types of formulation. In this study, we formulate this problem as a Linear Programming (LP) problem minimizing the mean dose in a buffer region, which is a region located around the tumour to ensure the effectiveness of the solution and the availability of the feasible solution. An illustrative plan of a treatment of cerebral tumour is given to demonstrate the effectiveness of the proposed method.

View full abstract

A magnetic stimulation system transfers stored energy on the capacitor bank to the stimulation coil through a power-supply-cable. A pair of two parallel cables has been used as the power-supply-cable. However, conventional parallel power-supply-cables (PPSC) have the following problems: 1. PPSCs generate high intensity leakage magnetic field around the cables (0.37T/7759A at z=4mm), and the leakage magnetic field may be the cause of EMI (electro magnetic interference) to medical equipment in the area. 2. PPSC generates high-intensity acoustic noise (85dB HL/7759A) by a repelling force between two conductors caused by high intensity pulsed current (around 10, 000A). Generated noise is enough to evoke AEP (auditory evoked potential) and contaminate the responses of magnetic stimulation. 3. High inductance of the PPSC decreases stimulus efficiency. We have introduced a coaxial cable power-supply-cable system (CPSC) to solve the above problems. The CPSC system decreases the leakage magnetic field and the acoustic noise to 1/80 and 18dB (HL) respectively. The CPSC system also improves stimulus intensity (dB_z/dt) by 43% compared with the conventional PPSC. The CPSC system allows the use of smaller diameter stimulation coils for more localized magnetic stimulation.

View full abstract

We recorded neuromagnetic fields from 4 Japanese subjects, mainly over the left temporal, parietal and occipital areas, responding to visual stimuli presented for 100 ms every 1.3-1.7s at lower left visual field. In two subjects the right hemisphere was also measured. The stimuli were randomly presented a kanji characters (50%) or a random dot patterns. (50%) The subject's task was to read silently kanji's. The following results were shown that kanji's as well as the random dot patterns activate (1) the medial occipital cortex at around 100ms, (2) the lateral occipital cortex at 150-180ms, and that kanji's activate (3) the left temporal cortex near the auditory area at 200-400ms. These results indicate that the kanji's as well as the random dot patterns activate first the medial occipital cortex and secondly the lateral occipital cortex. kanji's activate the left temporal cortex at 200-400ms as well. The fact that the random dot patterns did not activate the left temporal cortex suggests that the reading-related activity occurs in this region with the latency of 200-400ms.

View full abstract

The aim of the present study is to develop an advanced method for estimating the Hill's characteristic equation between force and velocity and to apply to detemination of the muscle force indices such as the maximal isometric force and maximal power. The estimation was performed according to the following procedures: 1) transforming the characteristic function to a linear form, 2) estimating the maximal isometric force providing the regression line with the maximal correlation coefficient using dynamic muscle force and velocity measured, 3) determining heat and energy liberation constant in the characteristic equation, which were obtained from the intercept of y-axis and slope of the regression line. In addition, the non-linear optimization was utilized for further improvement of the accuracy. In this meaning, the method is referred to as the Maximal Correlation Method. The theoretical validity of the estimation method was confirmed by the statistical simulation with a number of samples (10, 000 samples) from Gaussian random population. The statistical simulation and experimental results suggest that seven samples is the least number that allows accurate estimation of the muscle force indices with the estimation error of less than 10%. The present method appears to be a useful tool in investigating human dynamic movement.

View full abstract

In this study we have developed a transcutaneous digital communication technique by using Pulse-Code-Modulated (PCM) infrared rays. The communication protocol examined here is serial, asynchronous and half duplex. The electric power should be considered for a computer system implanted inside the body, because it is driven by batteries and the power is insufficient. Its size should also be miniaturized as small as possible. Therefore the following methods were adopted here; the weak infrared rays emitted inside the body were received by a high sensitive avalanche photodiode; the infrared rays of high brightness was used to send data into inside of the body; and the digital data were converted into an optical signal after the pulse-code-modulation (PCM). This modulation yields some advantages; 1) absorbing the response delay of the photosensor, 2) intensifying infrared rays luminescence outside body and 3) saving the electric power on the luminescence inside body. Some communication experiments were done by using pork with the skin; the communication speed was 9600 baud from inside to outside and 1200 baud in the opposite direction, 15 infrared LEDs were set for outside of the body, 4 for inside, three nickel hydride batteries of 1200mAh were used for an electric power inside the body. As a result, a stable bi-directional digital communication was realized through the skin and the pork of about 9cm and it could continue to operate for about 19 hours.

View full abstract

In this paper we describe a virtualized bronchus endoscope simulation system (VBE) which enables to observe inside the bronchus area from three dimensional (3-D) X-ray CT images. This system consists of two major functions (a) automated extraction of bronchus area, and (b) simulation of bronchus endoscope. First, the bronchus area is extracted by a 3D region growing algorithm with the parameters adjusted automatically. This extraction is performed fast by using a 3D painting algorithm. Therefore a user does not need to input bronchus area by hand for bronchus endoscope simulation. Next the endoscope simulation is done by visualizing the inside view of extracted bronchus area and changing the viewpoint and the view direction interactively. In this system, the user can observe any parts of bronchus from any directions, because the user can travel inside bronchus area freely by mouse operation. VBE has several advantages over the real endoscope. It is possible to observe the inside of very thin bronchi, observe without any pain, and measure feature values such as width and length. We applied this system to sixteen cases of real 3-D chest X-ray CT images which include tracheostenosis and tracheal tumor cases, and obtained good results. Furthermore we compared the real endoscope image and a VBE image, and showed the effectiveness of this system.

View full abstract

Spontaneous alignment of arterial endothelial cells was simulated using a three-dimensional computational fluid mechanical model of cultured endothelial cells. Endothelial cells were simulated using a 2D Gaussian distribution function, and placed on a flat plate with randomly assigned rotational movement. The Navier-Stokes equations of Newtonian fluid under steady flow conditions were solved using a finite volume method, and the absolute wall shear stress (WSS) at the summit of cells was calculated. Only the movement of cells which reduces the WSS was conserved under steady flow condition. The cell model eventually showed alignment after varying lengths of simulation time. This phenomenon was thought to simulate the flow-induced alignment of the endothelial cells in vivo.

View full abstract

Functional electrical stimulation (FES) has been developed in order to restore paralyzed motor functions of patients who have injured their upper motor neurons. Use of tongue movement was proposed as a control command for an FES system, because even quadriplegic patients can move their tongue voluntarily and generate much information easily. In this paper, we investigated the tongue movement by using a simple measurement system that simulated an input part of control commands in an FES system. The measurement system consisted of a 16-key membrane switch array, and a personal computer. The switch array detected the tongue movement as shift of touching area of the buccal region. We measured and discussed recognition rates by which the measurement system could recognize direction of tongue movement correctly. In this discussion, we assumed the following command-rules; transverse and longitudinal direction were accepted as a command, and reentry was required if oblique direction or undetected condition occurred. Five healthy subjects operated the system by their tongue in experiments, and the recognition rates were calculated. The rates within three entries were more than 85% even by less-experienced subjects; more than 95% by some subjects. The results of experiments during 5 days indicated that the repeatability of the recognition of the tongue movement was high. It implied that a precise calibration of a relative position between the system and a subject was not necessary. The results of this paper verified the fundamental feasibility of the use of tongue movement as a control command for an FES system.

View full abstract