Facial expression is very complicated and delicate, that is due to a mutual link among facial muscles under control of facial nerves. Generally, a front view of facial movement is symmetrical on right and left in healthy condition, however, it shows asymmetry in case that the neuro-muscular system is injured or disordered by the so-called facial paralysis or facial nerve palsy. In clinical diagnosis of the facial nerve palsy, quantitative and objective assessment of the facial movement is strongly required at present. In this study, a diagnosis system for the quantitative assessment of facial nerve palsy has been proposed using the optical flow method that presents the degree of precise movements in the facial area. Sequences of facial images have been analyzed to depict the optical flow for various kinds of facial expression. Right-left ratio of the facial movements obtained from the optical flow is very effective for the quantitative assessment of facial nerve palsy and it is available to the clinical use.

View full abstract

Mismatch negativity (MMN), a component of event-related potentials (ERP), was recorded to investigate the properties of temporal integration in auditory sensory memory, which integrates the short-term history of presented auditory temporal sequences to the sensory memory using a time window. At first, the MMN elicited by the infrequent change in a pair of closely spaced tones was evaluated to determine the temporal capacity of sensory memory. Under ignore condition, it was shown that the amplitude of elicited MMN decreased when the total duration of the tone pair (stimulus duration) increased, and for three out of four subjects, there was no significant MMN observed when the stimulus duration was 400ms. This result shows that the function of temporal integration existed in sensory memory and the maximum period of the time window was 300-400ms. However, MMN was also elicited under attend condition when the stimulus duration was 500ms, and a scalp topography of MMN showed additional activity in the frontal area. This might support the hypothesis of the contribution of higher-order processes to the sensory memory process. Secondly, the relationship between the temporal configuration of tone sequence and elicited MMN was evaluated under ignore condition. Temporal configuration was changed by inserting additional tones between the pair of tones. The MMN amplitude decreased when the number of inserted tones was increased, and the duration or temporal position of a single inserted tone affected the MMN amplitude. From the results of this experiment, it is believed that the capacity of the auditory sensory memory is influenced by the temporal configuration of the tone sequence to be stored, as well as its total duration. These results will help to approach the neuronal mechanism of auditory scene analysis.

View full abstract

The purpose of this study is to develop a computerized scheme that allows discrimination between benign and malignant clustered microcalcifications, thus aiding radiologists to better interpret mammograms. The fuzzy-logic based genetic algorithm (GA-Fuzzy) that we propose is applied for the classification of mammographic microcalcifications. As a preprocess, microcalcification regions are extracted from images using mathematical morphology and other means. Then, four feature values including the number, area, circularity and minimum distance of microcalcifications are calculated for classification using GA-Fuzzy. In the GA-Fuzzy process, Gaussian-distributed membership functions are determined from the means and standard deviations of the feature values calculated from training images. Subsequently, the genetic algorithm optimizes the membership functions, and training is completed. Next. the feature values calculated from unknown images are input into GA-Fuzzy system to classify the image. Our scheme was tested utilizing the database of the Mammographic Image Analysis Society (MIAS), which contains 13 benign and 12 malignant microcalcification cases. Of the images, there are ten of each benign and malignant cases respectively used for training. The remaining five cases were used for classification as unknown images. Various combinations of sets are employed to obtain the results. The average accuracy of the GA-Fuzzy technique was approximately 81% (sensitivity, 85%; specificity, 77%). These results show that our scheme can be regarded as a useful technique to classify microcalcification cases.

View full abstract

MRI angiography is useful for the diagnosis of blood vessels and blood fiow. The phase contrast method is suitable for quantitatively evaluating blood flow velocity because it is reflected in the phase. However, the velocity is low near the vessel wall, and the effect of the dephasing cannot be ignored in this region. On the other hand, the vessel wall is static and the effect of dephasing could be weak. The maximum velocity in the blood vessel also seems to be an accurate parameter because of the weak effect of dephasing. The static and maximum velocity regions therefore seem helpful for extraction of the vessel region, although the influence of the relaxation times and sequence parameters on dephasing have not yet been clarified. In this study we investigate the effects of these parameters on the measurement accuracy of the velocity and design a simple method for the regional extraction of blood vessels. The relaxation times, T_R, T_E and VENC, influenced the accuracy of velocity measurement. Although no dominant parameter was found among them, several combinations of parameters resulted in a significant difference in accuracy. The standard deviation of the phase in the static region increased with increasing relaxation time. We also extracted the popliteal artery region by growing the region until the phase reached a threshold value. This threshold value was set based on the standard deviation of the area in a static region. This method was found to be suitable for extracting the blood vessels from MR images obtained when using the phase contrast method.

View full abstract

An exchangeable microforceps for neurosurgery—which can be inserted into micromanipulators of a minimally invasive surgery-support system (called the hyper utility mechatronic assistant system, HUMAN)—was developed. This microforceps, which performs open-close operation at bending by means of a drive wire, is a single-opening type. It meets the specification of an outside diameter of 1mm and a maximum holding force of 1N, and a prototype was made of stainless steel by electric discharge machining. A microforceps drive-system was also constructed, and its dynamic characteristics were experimentally measured by a laser-sensor system. The forceps was then applied in an experiment on a swine brain tissue in vivo, and it was confirmed that its holding function is good enough for practical use. The results of these two experiments are being applied so that the holding-part mechanism and its dynamic characteristics can be further improved for clinical use. It is concluded that this newly developed microforceps will be agreat help in further developing minimally invasive neurosurgery using HUMAN.

View full abstract

Many diabetic patients carry a portable self-measuring blood glucose monitor, withdraw their own blood and examine their glucose level on a continuous basis. In order to support blood glucose control through the practical use of these data, database and network plans have been studied for the education, management and treatment of diabetes mellitus. In this study, we fabricated calorie calculation software that can calculate all food-intake calories automatically only by choosing the meal menu in mouse pointing device operation. It was possible for the data to be collected personally, even for elderly persons inexperienced in personal computer operation. In addition, multi-input information can be collected using portable commercial devices such as a blood glucose monitors and metabolic rate monitors. Next, on the basis of three data of blood glucose level, metabolic rate and food intake measured over approximately three months, the tendency for increase and decrease in blood glucose level was estimated based on data mining. As a result, it was shown that the accuracy of estimated variation was improved to 90%, although the error in estimating absolute value is yet large. However, understanding the lifestyles and habits of the patients and selecting appropriate input variables for individuals seems to be more important than extending the data collection period if the objective is to improve estimation accuracy. It was possible to partially clarify the effectiveness of data mining in blood glucose level estimation.

View full abstract

Deformation of the left ventricular wall during systole was analyzed using a magnetic resonance tagging technique.The subjects were two patients with dilated cardiomyopathy accompanied by a left bundle branch block. The circumferential strain at the short-axis section was employed as an index for evaluating cardiac contractility. The results showed that the circumferential strains in the septal walls of the patients were different from those of healthy humans, having positive values. This study proposes that circumferential strain could be an effective index for the quantitative evaluation of cardiac contractility.

View full abstract