Visual display terminal (VDT) syndrome causes dry eyes and related symptoms. In addition, the number of eye blinks is reduced in VDT workers. We have therefore developed a real-time eye blink counter system. This system employs face pattern recognition, eye-area detection, and tracking using a template-matching method for eye blink detection. The changes between successive frames in a movie captured by a web camera (30 fps, 320 × 240 pixels) were recorded. We then applied three different methods as image processing techniques for eye blink detection: the subtraction of successive frames, eye open/closed status detection by image scanning, and iris detection by template matching. The results showed that comparable performance could be achieved using all three methods. The system functioned well as a real-time eye blink counter and may also prove to be useful as a sleepiness or catnap detector, as well as find applications in man-machine interfaces and visual axis analysis.

View full abstract

There has been increasing interest in the diagnosis of breathing problems, including chronic obstructive pulmonary disease (COPD), in recent years. In hospitals, the patient's forced expiratory airflow is measured by spirometry to diagnose COPD and to monitor the effectiveness of treatment. However, the commercially available devices used in hospitals are contact-based systems, which may cause the patient to experience discomfort or pain, thus making it difficult to obtain accurate measurements. We have therefore developed a system employing a fiber grating three-dimensional sensor that displays breathing movements as a rough indicator of the patient's physical condition and a respiration monitor to measure body and breathing activity simultaneously. In the present study, we quantitatively analyzed the phase shift between the chest and abdomen seen in the respiratory movement wave patterns using a cross-correlation function. The results showed that patients with COPD exhibited a larger phase shift than healthy people. A significant difference was observed in the phase shift between healthy people and patients with COPD, and patients with more severe COPD showed a larger difference in the forced expiratory volume in one second percent (FEV1%) relative to healthy people. We conclude that this system is able to detect the characteristic changes in breathing pattern observed in patients with COPD and may prove to be useful as a new screening method for COPD.

View full abstract

In this report, we describe the development of a high-speed volume rendering system that combines 3D texture compression and parallel programming techniques for rendering multiple high-resolution 3D images obtained in medical or industrial CT. The use of 3D texture compression reduces memory consumption to 1/4 the original level with no negative impact on the display speed, and parallel processing using multiple cores in the CPU makes it possible to reduce the computational time with 3D texture compression to the same as that without 3D texture compression. By employing this approach, a personal computer with standard graphics capabilities can be used to display multiple high-resolution 3D images. In this evaluation, no large differences were observed between rendering images with and without 3D texture compression.

View full abstract

The Japanese Alzheimer's Disease Neuroimaging Initiative (J-ADNI), a large-scale multi-site clinical study to develop surrogate markers that lead directly to the essential process of Alzheimer's disease, was launched by All-Japan in Japan. Its goal is to develop methods for assessing the stages in the progression of Alzheimer's disease objectively and accurately through a variety of examinations, including clinical, psychological, imaging, and biochemical examinations. A variety of examinations are performed in the J-ADNI, but a large amount of money and a huge amount of effort are needed to gather such a massive amount of data and to control the data quality. To address this problem, the clinical test results must be acquired electronically and managed using a computer system. However, most existing computer systems are used only for the management of clinical findings and measurement values, and no computer systems manage not only these items but also MRI and PET image data and their imaging conditions. In this report, we propose a data checking system for the management of these items and for data quality control and describe the construction of a study management system to consolidate the management of the subjects' examination status. This will make it possible to gather the subjects' examination data appropriately and to analyze them in order to develop objective indicators. Our system manages clinical image data and the imaging protocols as well as the case report forms (CRFs) that have basically been managed by existing systems. Our findings suggest that our system will help to expand the range of applications of computer systems in large-scale multi-site clinical studies such as J-ADNI.

View full abstract

We propose new denoising and edge sharpening algorithms for medical ultrasound imaging. In ultrasound images, it is desirable to minimize speckle noise and enhance tissue boundaries (edges) while maintaining a natural appearance. The spatial frequency characteristics of the speckle patterns differ depending on the imaging conditions and scanning region, making it difficult to achieve good denoising performance for various types of images. To address this problem, we propose a denoising method employing multiresolution analysis that estimates the amount of noise in each spatial frequency band and adjusts the strength of denoising accordingly. We also propose an edge sharpening method employing a morphological filter that can avoid introducing artifacts. These techniques, when implemented in a special-purpose parallel processor, can achieve real-time display rates (60 frames per second).

View full abstract

Conventional intravital microscopy using a light microscope has been used to investigate the disease mechanisms of hypertension and diabetes at the level of the microcirculation unit. However, this method is limited to the assessment of surface vessels because only a thin layer is in focus while the remainder of the entire thickness is out of focus. On the other hand, X-ray intravital microscopy using a microangiography technique at SPring-8 can visualize both surface and penetrating vessels simultaneously. In in vivo experiments using diabetes model rats, diabetic vascular dysfunction in the small coronary arteries was visualized to evaluate the X-ray intravital microscopy system. At a high spatial resolution comparable to that of a conventional intravital microscope, the system has the ability to provide detailed anatomical information of the rat coronary arteries which is close to that of a human coronary angiogram. We have developed this X-ray intravital microscopy technique to visualize the surface and penetrating coronary arteries in rats as a new preclinical imaging system.

View full abstract

The aim of compressed sensing (CS) is to reconstruct signals and images from significantly fewer measurements than usual based on the sparsity in measured signals or images. In general, magnetic resonance images themselves are not sparse; however, they can be transformed to sparse signals by applying an appropriate function. We propose an improved CS method using a FREBAS transform as the sparsifying function rather than a wavelet transform or finite difference transform. Compressed sensing is an iterative reconstruction technique, so it is more computationally intensive than traditional inverse Fourier reconstruction. One barrier to the routine adoption of CS MRI is the delay between data acquisition and the reconstruction of acceptable images. In the present study, we attempted to accelerate CS reconstruction by employing a graphics processing unit (GPU) for general-purpose computing. The results showed that the use of the GPU dramatically accelerated CS reconstruction. A single image could be reconstructed in 1.3 s, indicating that the speed was increased by a factor of 9.

View full abstract

In FDG PET imaging, FDG uptake in tumors less than 2 cm in diameter is inherently underestimated due to the partial-volume effect (PVE) because PET images have limited spatial resolution. With regard to the deconvolution method, we use the Richardson-Lucy (RL) algorithm with total valuation (TV) regularization. In the first step of our PVE correction method, PET images are expanded with B-spline interpolation. In the next step, the RL algorithm with TV regularization is executed for an expanded image. In addition, the point spread function is assumed to be 3-D Gaussian. In the present study, our PVE correction method was evaluated using 12 spherical phantoms ranging from 6 to 40 mm in diameter. The results showed that our method reduced the inaccuracy due to the PVE in phantoms more than 8 mm in diameter and that the use of the RL algorithm with TV regularization provided values very close to the true SUV values for FDG uptake. Clinically, our method demonstrated an increase in the observed FDG uptake in a small brain tumor, resulting in improved contrast relative to the background.

View full abstract

We established the Global Collaboration Medical Network System Committee (GCM) in 2008, and since that time we have been involved in the development of a medical community system together with a medical association in the eastern part of Shizuoka prefecture. A large amount of research on the development of medical community systems has been conducted, and many approaches have been attempted in the past. However, since many such projects do not involve widespread collaboration and are based on client/server and web systems, various problems such as large initial investments and high running costs have hindered the popularization of global medical collaboration network systems. In addition, medical data and registration information for the nationwide resident registry network in Japan were lost due to the Great East Japan Earthquake, emphasizing the importance of preserving medical data in the event of natural disasters. To address the above problems, GCM has developed a cloud-based medical community system. We present a positive prototype system for a global collaboration medical system. The effectiveness of this system is demonstrated by the Lung Cancer Screening System.

View full abstract