Abstracts of Medical Imaging Technology of 2000.



January(Vol.18, No.1)

[-> return]
titleauthorpage
Technical Trend of Semiconductor Detector for Nuclear Medicine Nobuyuki NAKAMURA 3-8
Key words : CdZnTe, CdTe, Semiconductor, Radiation detector, Nuclear Medicine
Abstract : In recent years, the development of semiconductor detectors for nuclear medicine application has been underway in earnest. This significant technical innovation owes itself to the rapid progress of the manufacturing technology of semiconductor elements. Among these elements, the II-VI compounds CdZnTe and CdTe have been gaining special attention because it is possible to use them at the room temperature. For CdZnTe, the ut product of hole is relatively small, resulting in the deterioration of energy resolution from insufficient carrier collection. However, by revising the electrode structure, it is possible to cancel the holes and to collect almost all electrons, thereby making it possible to collect energy information correctly. For CdTe, it is possible to supply high-voltage and decrease leak current by using Schottky electrodes made of Indium. As a result, energy resolution and time response can be improved. Although problems such as a relatively large defective ratio and the lack of mass production techniques remain with CdZnTe and CdTe, other factors such as technical performance of the elements, assembly techniques and analog integrated-circuit technology are at levels such that production of new semiconductor detector systems for nuclear medicine can be realized.

[-> return]
titleauthorpage
SPECT with Acquisition 0f 511keV Gamma Photon Masatoshi TAGUCHI 9-14
Key words : SPECT, Coincidence, 18F-FDG, Transmission CT
Abstract : 511keV SPECT imaging using 18F-FDG has a highly specific detectability for tumors. The coincidence imaging method principally provides a collimator-less SPECT with a high system sensitivity and is resultantly expected to provide a better image quality. 511keV SPECT has a possibility to become available widely if delivery system of 18F-FDG are well organized making use of a feature that 18F has a comparably longer half life of 110 min. Fundamental principle and improvement of 511keV SPECT imaging and its possibility in the future are briefly described below:

[-> return]
titleauthorpage
Radiation detectors in Positron Emission Tomography Hideo MURAYAMA 15-23
Key words : Positron emission tomography, Nuclear medicine, Radiation detector, scintillator, Position sensitive detector
Abstract : Current status of photon detectors in positron emission tomography (PET) is reviewed. The recent development of a three-dimensional (3D) data acquisition technique has had a significant impact on the design and operation of PET scanner. The development of position encoding detectors with the depth-of-interaction (DOI) aims for a restoration of the spatial resolution uniformity and increased sensitivity through a reduction of the tomograph ring radius, and the practical ones are under developing to improve the image quality in the image reconstruction for 3D PET. Combined with high light yield scintillators, new avalanche photodiodes (APDs) show many advantages over photomultiplier tubes and PIN photodiodes, such as their compactness, low noise and potential mass production.

[-> return]
titleauthorpage
Scatter, Attenuation and Detector Response Correction of SPECT Hiroyuki SHINOHARA, Tomoaki YAMAMOTO, Hideharu SUGIMOTO, Takeyuki HASHIMOTO, Munehiro TAKAHASHI, Takashi YOKOI
Key words : SPECT, Scatter correction, Attenuation correction, Detector response correction
Abstract : Scatter, attenuation and detector response in single photon emission computed tomography (SPECT) imaging contribute to the error in quantification of activity concentration and decrease in contrast of reconstructed images. Correction methods for these three factors are introduced and related physical phantom studies are reported.

[-> return]
titleauthorpage
Analytical Methods of Three-Dimensional Image Reconstruction Eiichi TANAKA 33-39
Key words : 3D PET, Image reconstruction, Reprojection method, Rebinning methods, Cone-beam SPECT, Rotating slant-hole SPECT
Abstract : This paper describes the theory and practice of analytical methods of three-dimensional (3D) image reconstruction for PET and SPECT. In 3D PET, the 3D image is reconstructed from parallel-beam projections taken at various slant angles. The main topics are "3D filtered backprojection method" for a constant acceptance angle of projections, "3D reprojection method" used in practical 3D PET scanners, and "rebinning methods" as the fast approximate reconstruction methods. This paper also describes a sufficiency condition for 3D image reconstruction from cone-beam projection data obtained with cone-beam SPECT, and gives a fundamental reconstruction method on the basis of Radon transform and some practical algorithms. Principle of rotating slant-hole SPECT and its recent topics are mentioned briefly.

[-> return]
titleauthorpage
Fast Iterative Methods for Image Reconstruction in SPECT/PET Hiroyuki KUDO 40-45
Key words : SPECT, PET, Image reconstruction, Iterative method
Abstract : Iterative image reconstruction methods in SPECT and PET possess significant advantages over analytical reconstruction methods. This paper reviews recently developed three fast iterative methods (Fourier preconditioned conjugate gradient, block iterative, and coordinate descent methods).

[-> return]
titleauthorpage
Metacomputing Environment for Magnetoencephalography (MEG) Susumu DATE, Yuko MIZUNO-MATSUMOTO, Shinichi TAMURA, Yoshinobu SATO, Reza A. ZOROOFI, Yuji TABUCHI, Shinji SHIMOJO, Youki KADOBAYASHI, Haruyuki TATSUMI, Hiroki NOGAWA, Kazuhiro SHINOSAKI, Masatoshi TAKEDA, Tsuyoshi INOUYE, Hideo MIYAHARA
Key words : Networking, MEG, Brain function, GLOBUS, MPI, Parallel computing
Abstract : In this paper, we propose a method for evaluating brain function using magnetoencephalography (MEG) in a metacomputing environment, that is, a widely distributed computing environment. For the metacomputing environment, we adapt the GLOBUS system, which provides various software components for this environment, and its API (Application Programming Interface) for MPI (Message Passing Interface), called MPICH-G. Through the use of this metacomputing environment, three processes (acquisition of data from MEG, analysis, and visualization) are smoothly integrated into a single seamless process on the wide area network. As a result, the total time and cost of reaching the final diagnosis are reduced. We may also be able to geometrically distribute these three processes in the future, which would support remote diagnosis. In addition, with the use of MPI, analysis can be performed by parallel processing. We constructed a small-scale system on the local area network and implemented analytical processing on a massively parallel supercomputer, Exemplar. Our results showed that we could achieve an 8-fold increase in parallelism in the analytical processing, and that our graphical interface for MEG should be very helpful in realizing an efficient diagnostic system for brain function analysis.

[-> return]
titleauthorpage
Propagation of Epileptiform Discharges Using Wavelet-crosscorrelation Analysis in MEG Yuko MIZUNO-MATSUMOTO, Shinichi TAMURA, Yoshinobu SATO, Reza A. ZOROOFI, Susumu DATE, Yuji TABUCHI, Kazuhiro SHINOSAKI, Satoshi UKAI, Ryohei ISHII, Tsuyoshi INOUYE, Haroyuki TATSUMI, Youki KADOBAYASHI, Shinji SHIMOJO, Masatoshi TAKEDA, Hideo MIYAHARA
Key words : Wavelet-crosscorrelation analysis, Non-stationary analysis, Epilepsy, Propagation, Localization, Magnetoencephalography (MEG)
Abstract : Wavelet-crosscorrelation analysis, a new non-stationary method, is proposed as a technique for revealing the locations of epileptic foci and the propagation of epileptiform discharges. A patient with simple partial seizures was examined using whole-head magnetoencephalography (MEG). The time-lags between all sensor sets were calculated to obtain the maximum wavelet-crosscorrelation coefficient. The results showed that this method is useful for evaluating the epileptogenesis of epileptic phenomena seen in MEG and EEG studies. This method can be used to determine the propagation time and the correlation of two non-stationary waves containing various frequency components.

[-> return]
titleauthorpage
Improved Registration Method for Correcting Local shifting on DSA images Kyoichi HIROSHIMA, Junji NISHINO, Tomohiro ODAKA, Hisakazu OGURA, Testuya FUKUSHIMA, Yasuhiro NISHIMOTO, Masato TANAKA, Hiroyuki KOMURO, Yasushi ISHII 71-83
Key words : Digital subtraction angiography (DSA), Misregistration, Registration, Motion artifact
Abstract : We have investigated digital subtraction angiography (DSA) images, which are used to extract blood vessels. In this paper, we propose a new algorithm for eliminating motion artifacts using a local movement method over the image. The proposed method was found to correct artifacts in the images extremely well. Peripheral blood vessels around the nasal cavity and/or the orbit, which could not be observed well by conventional registration methods, were extracted clearly by the proposed method.

[-> return]
titleauthorpage
Comparison of Iterative Image Reconstruction Methods in Single Photon Emission CT Hiroshi URABE, Kotoko MORIKAWA, Koichi OGAWA 84-93
Key words : Medical imaging, SPECT, Image reconstruction, Iterative methods
Abstract : In this paper, we compare the quality of images reconstructed by ART (algebraic reconstruction technique) and SIRT (simultaneous reconstruction technique ), GRADY ( gradient method ), CONGR ( conjugate gradient method ), ML-EM ( maximum likelihood - expectation maximization ), and MAP-EM ( maximum a posteriori - EM ),which are known iterative image reconstruction methods.





March(Vol.18, No.2)

[-> return]
titleauthorpage
Flat Panel Displays for Diagnosis and Monitoring of Medical Imaging Hiroshi TSUTSUI 101-109
Key words : Flat Panel Display (FPD), Liquid Crystal Display (LCD), Plasma Display Panel (PDP), Radiography, Medical imaging
Abstract : Recently almost of visual data obtained by the modalities containing CR, x-ray CT, MRI and Ultrasound have been treated digitally, and the diagnosis using CRTs and the tele-radiography have been utilized. On the other hand, the technologies of the flat panel displays (FPD) like the Liquid Crystal Display (LCD), the Plasma Display Panel (PDP) and so on, are improved and performances of these displays become to be acceptable for the diagnostic and the clinical use in the past few years. In this paper, the mechanism, the performance and the potential of the FPDs for medical applications are described.

[-> return]
titleauthorpage
CRT Diagnosis -Observer's Performance- Takeo ISHIGAKI 110-114
Key words : CRT diagnosis, Computed Radiography, Observer's Performance
Abstract : The diagnostic efficacy of hard-and soft-copy computed radiographic (CR) images of the chest in the detection of subtle pulmonary abnormalities was discussed in comparison of conventional screen-film images. ROC studies showed no statistically significant differences among hard-copy screen-film images, hard-copy CR images, and soft-copy CR images with both monochromatic and color monitors. Soft-copy chest CR images are acceptable and available in place of screen-film images for primary interpretation of subtle interstitial lung diseases.

[-> return]
titleauthorpage
Evaluation of CRT Diagnosis from Clinical Point of View Shinichi HORI¡¤ Kazuo AWAI¡¤Nobuko EGUCHI¡¤Kumiko HIRAISHI 115-119
Key words : Diagnostic images, Diagnosis, CRT, PACS
Abstract : Recent advancement of imaging technology is going to change the diagnostic radiology by producing many sort of imaging techniques and large number of images. It becomes difficult to observe a large amount of images by the conventional method using hard copies. Picture archiving and communication system (PACS) can solve current problems in management of diagnostic imaging department. The new image observation method, CRT diagnosis was compared with conventional film method in daily clinical works. The various CRT functions helped us to observe many images in a short time. Diagnostic information was well utilized by changing the image condition by diagnostic radiologist. The selected images with diagnostic report were simultaneously transferred to the hospital information system.

[-> return]
titleauthorpage
Application of Surgical Simulation and Navigation System with 3D Imaging Kenshi KANEKO, Yasushi FURUTA, Nobuaki ISHII, Chihiro TAKAHASHI, Satoshi TERAE, Kazuo MIYASAKA 121-126
Key words : Computer-assisted Surgery, Navigation system
Abstract : Surgical navigation and simulation systems with 3D imaging allows operations to be more precise. The advances in diagnostic imaging, especially in the fields of computed tomography (CT) and magnetic resonance imaging (MRI) and improvements in data processing technology enable real time interactive surgical localization with great accuracy. We present the composition of this system, the method of processing preoperative 3D-data, correlating this data with an actual surgical space, and discuss clinical applications, mainly in neurosurgery.

[-> return]
titleauthorpage
Realtime 3D-DSA and CRT diagnosis Toshiyuki NAKAMURA 127-130
Key words : Angiography, 3D-DSA, CRT diagnosis
Abstract : We describe a real time Three-Dimensional(3D) DSA(digital subtraction angiography) with CRT(cathod ray tube) diagnosis in this paper. We use rotation DSA and precession DSA as the 3D DSA in additional to conventional DSA in clinical examination. Real time 3D DSA is very useful to describe a fine lesion, which make the number of the study¡¡increasing in our hospital. CRT diagnosis is neccessary to interprete¡¡the motional information, which enhance the information of 3D DSA.

[-> return]
titleauthorpage
Medical and Economic Requirements for the Promotion of Filmless Radiology Using CRT Display Chikazumi KURODA Yoshiyuki FUDEMOTO 131-135
Key words : CRT, Filmless radiology, Teleradiology
Abstract : The medical and economic advantages of filmless radiology using a CRT display in the clinical setting are discussed. The popularity of PACS and teleradiology is an excellent context in which to promote these advantages of filmless radiology. The major medical benefit of filmless radiology is improved accessibility of electronic clinical records and images. In addition, the economic benefits of filmless radiology will contribute to the improvement of the Japanese medical economy.

[-> return]
titleauthorpage
Detailed Extraction Method for the Lung Region Using the Active Cylinder Model in Chest X-ray CT Images Toshiaki OKUMURA, Shinji YAMAMOTO, Tohru MATSUMOTO, Yukio TATENO, Takeshi IINUMA, Mitsuomi MATSUMOTO 136-148
Key words :Image processing, Region extraction, Active Cylinder Model, Chest X-ray CT images
Abstract : We are developing an automatic system for recognizing lung cancer shadows in chest X-ray CT images to support lung cancer mass screening. The development of an extraction method for the lung region in chest X-ray CT images is an important preprocessing method for the above computer-aided diagnosis system. We have already proposed a model-based rough extraction method for the lung region using a typical patient sample as a model. This method involves the affine transformation of the model to fit it to the input image, and the transformed model image is defined as the rough lung region for the input image. In this paper, we propose a detailed extraction method using the Active Cylinder Model. The results of the rough extraction method are used as initial control points in the Active Cylinder Model. The control points are moved based on the internal energy and the image energy. The internal energy is position information of the adjoining control points, and the image energy is image information such as edge information calculated from the input image. This method was applied to 68 sample chest CT images, and was found to be very effective and practical.

[-> return]
titleauthorpage
Image Management System for Integrating Distributed Servers with a Web Browser in a Radiology Department Masato TANAKA, Hideki UESAKA, Yasutaka KAWAMURA, Hiroyuki KOMURO, Harumi ITO, Yasushi ISHII, Yoshiharu YONEKURA, Norihiro SADATO, Hisakazu OGURA 149-157
Key words : Masato TANAKA, Hideki UESAKA, Yasutaka KAWAMURA, Hiroyuki KOMURO, Harumi ITO, Yasushi ISHII, Yoshiharu YONEKURA, Norihiro SADATO, Hisakazu OGURA
Abstract : Sixteen years ago, we introduced a medical image management system using microfilms at the Department of Radiology of Fukui Medical University Hospital. This system is very useful for clinical education and research. However, it has some significant shortcomings, such as a decrease in image quality, loss of microfilms, high running costs, and insufficient storage space. We have developed an electronic medical image management system based on distributed data resources which we plan to use to expend and complement our existing microfilm-based system. Medical image acquisition is performed automatically when an operator sends images to the laser imager, and the captured images are sent to a server called the Image Server using the DICOM protocol. The Image Server compresses the images using the JPEG method and prepares two types of images: a full image with the same number of pixels as the original image and an abbreviated image with one half the number of pixels. Then, the Image Server sends the attributes of the images to a Total Control Server. The Total Control Server maintains and updates an index file for searching, and integrates the distributed resources. Client computers sited at the clinical offices search the results of examination by communicating with the Total Control Server and display the images received from the respective image servers directly using a Java applet executed on a web browser. We have used this system in actual practice for 10 months and have stored 150 million full and abbreviated images (total disk volume: 12.1 GB). This electronic system is able to maintain high-speed response due to the distributed resources and the reduction in image volume, and has become indispensable in ensuring smoothly examinations and routine diagnostic work.

[-> return]
titleauthorpage
Elasticity Analysis by MR Elastography using the Instantaneous Frequency Method Osamu OSHIRO, Mikio SUGA, Kotaro MINATO, Jun OKAMOTO, Osamu TAKIZAWA, Tetsuya MATSUDA, Masaru KOMORI, Takashi TAKAHASHI 158-166
Key words : Elasticity, MRI, MR Elastography, Instantaneous Frequency Method
Abstract : This paper describes a calculation method for estimating the elasticity of human organs using magnetic resonance elastography (MRE) images. The method is based on the instantaneous frequency method, which is very sensitive to noise. Therefore, the proposed method also incorporates a noise-reduction function. In the instantaneous frequency method, Fourier transform is applied to the measurement signal. Then, inverse Fourier transform is performed after the negative frequency component is set to zero. In the proposed method, noise is reduced by processing in which the positive higher frequency component is also set to zero before inverse Fourier transform is performed. First, we conducted a simulation study and confirmed the applicability of this method and the noise reduction function. Next, we carried out a phantom experiment and demonstrated that elasticity images could be generated, with the gray level corresponding to the local frequency in MRE images.





May(Vol.18, No.3)

[-> return]
titleauthorpage
Virtual Imaging of the Lung Based on Computational Morphology Hiroko KITAOKA, Shinich TAMURA 173-178
Key words : Computer, Morphology, Virtual imaging, 3D model, Simulation, Lung
Abstract : We propose an entirely new approach named "virtual imaging of an organ based on computational morphology". The computational morphology includes the structural model of the organ and mathematical models of the diseases, and generates a virtual organ in computer. The virtual organ is then divided into cubic voxels with the side of 0.5 mm, and virtual CT value for each voxel is calculated according to the tissue property within the voxel. The validity of the model is examined by comparing the virtual images with the clinical ones. Based on the models validated, image analysis methods for diagnosis are developed.

[-> return]
titleauthorpage
Fractal Analysis of Emphysema in X-ray CT and Simulation Michiaki MISHIMA 179-186
Key words : COPD, LAA, Terminal airspace, Diffusing capacity
Abstract : Increases in the low attenuation areas (LAA) of chest x-ray CT computed tomography images in patients with chronic pulmonary disease (COPD) have been reported to reflect the development of pathological emphysema. We examined the statistical properties of LAA clusters in COPD patients and normal subjects. In COPD patients, the percentage of the lung field occupied by LAAs (LAA%) ranged form 2.6 to 67.6. In contrast, LAA% was always <30% in healthy subjects. The cumulative size distribution of the LAA clusters followed a power law characterized by an exponent D. We saw that D is a measure of the complexity of the terminal airspace geometry. The COPD patients with normal LAA% had significantly smaller D values than the healthy subjects, and the D values correlate with the diffusing capacity of the lung. We interpret there results by using a large elastic spring network model and found that the neighboring smaller LAA clusters tend to coalesce and form larger clusters as the weak elastic fibers separating them break under tension. This process leaves LAA% unchanged whereas it decreases the number of small clusters, which results in a reduction in D similar to that observed in early emphysema patients. These findings suggest that D is a sensitive and powerful parameter for the detection of the terminal airspace enlargement that i occurs in the early emphysema.

[-> return]
titleauthorpage
Mapping of Pulmonary Ventilation with Nonradioactive Xenon-enhanced CT Yoshifumi YASUHARA, Junpei IKEZOE, Kenji SHIMIZU 187-192
Key words : Thorax CT, lung ventilation, Xenon
Abstract : Nonradioactive xenon gas has been used as a contrast medium for the evaluation of pulmonary ventilation because the attenuation of the lung is proportional to the alveolar xenon-gas concentration. Since, xenon can be detected in the lung with computed tomography (CT) imaging, the areas with impaired ventilation can be demonstrated on CT image. We measured the pulmonary enhancement accurately by scanning the lungs at the same degree of inflation before and after inhalation of xenon gas with a continuous rotation scanner. With this technique we could create a detailed map of pulmonary ventilation. The relation between the area with decreased ventilation and the lesion on CT image was easily understood with the ventilation map from nonradioactive xenon-enhanced CT.

[-> return]
titleauthorpage
An Extraction Method of the Pulmonary Blood Vessel Based on Thin-section 3D CT Image Noboru NIKI, Yoshiki KAWATA 193-197
Key words : 3D-CT image, Pulmonary blood vessel, Extraction method
Abstract : This paper presents the extraction methods of the pulmonary blood vessel based on thin-section 3D CT images. These methods are evaluated using clinical data. The effectiveness of these results is shown.

[-> return]
titleauthorpage
Successive Approximation Algorithm for Cancellation of Artifacts in DSA Images Raiko FUNAKAMI, Kyoichi HIROSHIMA, Junji NISHINO, Tomohiro ODAKA, Hisakazu OGURA, Tetsuya FUKUSHIMA, Yasuhiro NISHIMOTO, Masato TANAKA, Harumi ITO, Kazutaka YAMAMOTO 199-206
Key words : Digital subtraction angiography (DSA), Misregistration, Registration, Motion artifact
Abstract : In this paper, we propose an algorithm for cancellation of artifacts in DSA images. We have already proposed an automatic registration method based on the detection of local movements. When motion of the object is large, it is difficult to estimate the exact movement, and the cancellation of artifacts may therefore fail. The algorithm we propose here is based on a simple rigid model. We present the results of applying the proposed method to a series of experimental X-ray images, as well as the results of applying the algorithm as preprocessing for a registration method based on local movement.

[-> return]
titleauthorpage
Analysis of Morphology of Pulmonary Nodules Using a Fourier Descriptor Mitsuru KOMIZU, Hisayasu MATSUO, Yoshio YANAGIHARA, Kiyoshi MURATA, Masashi TAKAHASHI, Satoru MATSUO, Hiromitsu HAMA 207-216
Key words : Fourier descriptor, Pulmonary nodule on CT image, Morphological feature, Computer aided diagnosis
Abstract : The usefulness of computer aided diagnosis (CAD) systems in the differential diagnosis of pulmonary nodules detected by lung cancer screening CT has been widely acknowledged. In this study, we quantified the morphological features of such nodules and assessed whether they could be used for objective evaluation to improve diagnostic accuracy, based on hypotheses regarding differences in growth appearance between benign and malignant nodules. Quantification was achieved for both high-resolution CT images and simulation images using Fourier description. Benign pulmonary nodules ( inflammatory scars) showed high amplitude in the lower harmonics, while malignant nodules (lung cancers) showed high amplitude in the higher harmonics. These values were compared by Mahalanobis generalized distance multivariate analysis. The mean predictive values of this analysis for the inflammatory scar and lung cancer groups were 85% and 90%, respectively. The mean predictive value for the two groups was 87.5%.
Correspondence: E-mail me

[-> return]
titleauthorpage
Fuzzy Computation for Detecting Edges of Low-Contrast Substances in Urinary Sediment Images Ning-Feng ZENG, Keiji TANIGUCHI, Sadakazu WATANABE, Yutaka NAKANO, Hiroyuki NAKAMOTO 217-229
Key words : Edge detection, Low-contrast substance, Urinary sediment image, Fuzzy reasoning, Edge amelioration
Abstract : This paper proposes a robust model for detecting the edges of low-contrast substances in urinary sediment images. The edge detection process is divided into four stages: denoising, scaling, detection, and amelioration. In the first stage, the noise is reduced using optimum local statistics. In the second stage, the intensity is classified into different scales, and our proposed algorithm is then used to detect the edge of each scale. In the third stage, the edge is detected based on both the local pixel and the characteristics of neighboring pixels. It is difficult to distinguish between the intensity of low-contrast substances and that of the background, but local variance is much more easily distinguished. Edgeness (edge sharpness ) is evaluated by applying the fuzzy membership function of the variance and other fuzzy membership functions which are based on local pixel characteristics. In the final stage, although most of the edge segments are detected perfectly, some parts may remain ambiguous if noise is present. In this case, some of the detected edges that are false segments can be ameliorated using certain local features of the object based on the overall view. The experimental results show that our scheme is quite robust. This method is capable of detecting the closed edges of low-contrast substances with accuracy comparable to that of the human eye.

[-> return]
titleauthorpage
Morphological Method for Automatic Extraction of the Coronary Arteries Ying QIAN, Shigeru EIHO 231-239
Key words : Morphological operation, Coronary artery, Image processing
Abstract : Coronary arteriography is a clinically important diagnostic tool for the evaluation of coronary artery disease, and can provide detailed information. For the quantitative assessment of the coronary arteries, it is necessary to extract the edges of the arteries from coronary arteriograms. Several studies concerning the extraction of vessel edges have been published, and automatic extraction of vessel edges has been used in clinical diagnostic systems. However, these methods are not satisfactory, because manual modification by the operator is unavoidable in some cases. To reduce manual operation, accurate and automatic extraction of the coronary arteries is necessary. In this paper, we propose a new technique for automatic extraction of the coronary arteries using morphological operators. This method includes the following steps: 1) contrast enhancement using a morphological Top-Hat operator, 2) enhancement of thin vessels and reduction of pulse noise using a morphological erosion operator, 3) elimination of obvious background pixels by semi-binary thresholding, and 4) extraction of the coronary arteries by labeling and counting the area.

[-> return]
titleauthorpage
Evaluation of Diaphragmatic Movement during Full-Inspiration Breath-Holding Toshizo KATSUDA, Shigeru EIHO, Chikazumi KURODA 240-244
Key words : Full-inspiration breath-holding, Lung imaging, Diaphragmatic movement and movement direction
Abstract : In this article, we describe diaphragmatic movement during full-inspiration breath-holding. Diaphragmatic movement was analyzed on angiographic films during 24-second full-inspiration breath-holding in 36 patients. The distance between the liver apex and the upper edge of the field was measured on 15 films in each patient. The direction of movement was also analyzed. In more than 60% of patients, the diaphragm moved in the cranial direction during long-duration full-inspiration breath-holding. In lung imaging, scanning should be performed from base to apex because diaphragmatic movement is more often in the cranial direction.





September(Vol.18, No.5)

[-> return]
titleauthorpage
Diagnosis Systems of Valvular Heart Disease Using the Concept of Prototypicalness of Patients Toshikazu TOBI, Jiro ANBE 671-680
Key words : Fuzzy set theory¡¤Diagnosis method¡¤Valvular heart disease
Abstract : We have applied the diagnosis method proposed by D.Norris, B.W.Pilsworth and J.F.Baldwin to the diagnosis of valvular heart diseases. This method uses concepts from fuzzy set theory. We performed the experiments in order to evaluate effectiveness of the proposed diagnosis method. Also, we extended the original method to handle partial manifestation of symptoms and severity of diseases by using fuzzy sets. In addition, we introduced the concept of prototypicalness of patients with a particular disease to improve the performance of the diagnosis. The results of the experiments are very promising for the improved diagnosis method. In the best case, we achieved a true percentage rates of 89%. In this paper, we report the quantitative results of the experiments.

[-> return]
titleauthorpage
Fuzzy Information Granulation on Three-Dimensional Medical Image Processing Yutaka HATA, Syoji KOBASHI, Yuri KITAMURA, Toshio YANAGIDA 681-687
Key words : Fuzzy information granulation, Medical image segmentation, Magnetic resonance imaging, Magnetic resonance angiography, Human brain, Blood vessel
Abstract : This paper describes an application of fuzzy information granulation (fuzzy IG) to medical image segmentation problems. The issues are whole brain extraction from human brain magnetic resonance (MR) images, and blood vessel extraction from three-dimensional time-of-flight (TOF) MR angiography images. These methods are based on both the fuzzy IG concept and physician's knowledge represented by fuzzy linguistics.

[-> return]
titleauthorpage
Prediction of Resultant Profile After Orthognathic Surgery Employing Neo Fuzzy Neuron Noriaki SUETAKE, Ichiro MASUI, Takeshi YAMAKAWA 688-697
Key words : Neo fuzzy neuron, Orthognathic surgery, Prediction, Profile
Abstract : During presurgical planning of orthognathic surgery, clinicians predict a resultant facial profile which is much influential in facial esthetics. Many researches to predict a postoperative profile were reported, most of which used linear regression analyses with changing value of profile landmarks and skeletal landmarks. But the prediction employing changing values of the landmarks are deficient in information of face size and thickness of the soft tissue. Also, a profile change is a non-linear result from the interaction of not only skeletal movement but also soft tissue thickness or tonicity. This paper proposes an innovative prediction method by the neo fuzzy neuron. Materials used in this study were lateral cephalograms taken before and after surgery of 38 patients had bimaxillary orthognathic surgery. The skeletal and facial landmarks were measured two-dimensionally. The preoperative X-Y coordinates of hard and soft tissue landmarks were input variables; postoperative coordinates of soft tissue landmarks were output. Comparing prediction errors between the neo fuzzy neuron method and the linear multiple regression method, the neo fuzzy neuron method was most suitable for prediction of resultant profile after orthognathic surgery.

[-> return]
titleauthorpage
Application of Neural Networks for Computer-Aided Diagnosis Toshiyuki KAWAZU, Kazuyuki ARAKI, Shigenobu KANDA 698-702
Key words : Computer-Aided Diagnosis, Neural Network, Metastatic lymph node
Abstract : Neural networks are computing architectures inspired by nervous system. It can be used in many kinds of aspects even when formal analysis would be difficult or impossible, such as pattern recognition or pattern classification. Recently, the studies of neural networks for Computer-Aided Diagnosis(CAD) in medical science have been well documented. In this article we evaluated the diagnostic accuracy in differentiating metastatic cervical lymph nodes from benign lymphadenopathy using neural networks. We used 100 lymph nodes (50 were metastatic and 50 benign lymph nodes). The network was trained with three layer feed-forward backpropagation. Diagnostic accuracy obtained in this study was much the same as that of clinicians. Neural network seemed to be useful for differential diagnosis.

[-> return]
titleauthorpage
Immunity-Based Systems¡§A Specification and Applications Yoshiteru ISHIDA 703-708
Key words : Immunity-based systems, Signal processing, Self/non-self discrimitnation, Noise filtering, Complex system
Abstract : A unique feature of the information processing by the immune system is so called¡Èself/non-self discrimination¡É. The immune system carries out the task in the difficult situation where the non-self (antigen) cannot be predicted beforehand. The immunity-based system which mimic the information processing of the immune system may be realized potentially in many domains where the self (non-self) can be conceptually captured. This paper explores a specification of immunity-based systems by presenting two examples of applications to signal processing. Autonomous unit is introduced as a primitive construct of the immunity-based system. Similarly to the lymphocytes (the cell of the immune system that determine the specificity of the immunity), the unit can not only recognize but react against a specified feature of the signal.

[-> return]
titleauthorpage
Artificial Immune Network for Feature Extraction of Patterns Toshiyuki SHIMOOKA, Yasufumi KIKUCHI, Koichi SHIMIZU 709-714
Key words : Pattern recognition, Feature extraction, Immune system, Idiotypic network
Abstract : The application of an idiotypic network model to the feature extraction of patterns was attemptted. When we assume an observed pattern as an antigen, we can consider the immune system as the pattern-matching system using antibodies as partial templates. The antigen can be represented by steady-state solution that is the internal image resulted from the interaction of antibodies. Using the Farmer's mathematical model, the dynamics of simple cyclic networks were studied. It was found that we could classify observed patterns by analyzing steady-state solutions. Through this study, the usefulness of the idiotypic network for feature extraction of patterns was verified.

[-> return]
titleauthorpage
Fractal Dimension Texture Analysis of Brain Function Images in Diffuse Encephalopathy Koji UEMURA, Hinako TOYAMA, Kenji ISHII, Akihiko UCHIYAMA 715-723
Key words : Fractal dimension, Box count method, Diffuse encephalopathy, Cortical surface, Cerebral blood flow, Glucose metabolism
Abstract : In toxic or infectious encephalopathy, PET images of cerebral blood flow and the cerebral metabolic ratio of glucose (CBF and CMRGlc, respectively) are abnormally heterogeneous across the cerebral cortex with a speckled appearance. In this study, we evaluated the nature of the heterogeneity of 3D-cortical surface images quantitatively by fractal dimension texture analysis. The 3D-brain surface was projected onto a 2D-plane by Mollweide projection. Then, after smoothing the PET images of the projected brain surface an appropriate number of times, the mean fractal dimension was estimated by the box count method. The fractal dimension reflected the grain size of the speckle distribution, which differed from the absolute values of CBF and CMRGlc. The fractal dimension values in CBF images were higher than those in CMRGlc images. In patients with brain disease, the fr actal dimension values were lower than those in normal subjects in CBF images. The results suggest that this method allows quantitative measurement of the cortical surface texture in a 2D plane and is useful for analyzing PET and SPECT data of patients with diffuse encephalopathy.

[-> return]
titleauthorpage
System for Selecting and Generating Images Controlled by Eye Movements Applicable to CT Image Display Yoshio YANAGIHARA, Hiromitsu HAMA 725-733
Key words : Eye control application, CT image selection mechanism, Eye gaze measurement, Medical image display system, Computer-assisted system
Abstract : Several methods for controlling computer systems by eye movements have been studied recently as user interfaces for computers. On the other hand, overlaying medical images on the patient will lead to the development of new diagnostic imaging techniques. When the is performing other tasks simultaneously, gaze control is a useful interface for the display system. We save developed a new system for selecting CT images by eye movements. The display area is set near the center of the visual field and the control areas are arranged near the edges and in the corners. As there is some time lag between the acquisition of eye-movement data and the generation of display images, the system includes multiple specialized computers for each process. To avoid the small validation of eye movements, a hysteresis filter is employed. Experiments showed that the filter eliminated validation more effectively than a moving-average filter. The processing time to display selected images was 1.6 seconds for three images and 4.0 seconds for fifteen, including the waiting time for effective control (1 sec).
Correspondence: E-mail me

[-> return]
titleauthorpage
Evaluation of Accuracy of Localizing an Intravascular Ultrasound Probe for Endoscope Navigation Masatsugu HIRANO, Bo YANG, Yoshinobu SATO, Shinichi TAMURA, Yoshifumi NARUMI, Munehiro MAEDA, Hironobu NAKAMURA 735-739
Key words : Ultrasound image, Position marker, 3D localization, Endoscope, Internal navigation
Abstract : We describe a method for identifying the 3D position and direction of an ultrasound scanning probe introduced into the body to obtain endoscopic images. A mark obtained by an ultrasound pulse from an external transducer is shown on the endscopic image. The position of the mark (direction and distance between the scanning probe inside the body and the marker transducer outside the body) is extracted and measured on the scanned image. Then, the absolute position of the ultrasound probe inside the body is determined by identifying the absolute position of the external transducer with an optical localizer. We performed experiments using phantoms several centimeters in size. The water in the water tack was set to various temperatures. Actual position data (ground truth) was obtained by scanning the ultrasound probe with the pen probe of the optical localizer. Experiments for determining the 3D position of the probe were performed at 18 points in the water tank. The average error was 3.13¡Þ1.13 mm.

[-> return]
titleauthorpage
Interactive MR Imaging Technique: Initial Experience Kenji SHIMIZU, Shigenori SUGATA, Teruhito MOCHIZUKI, Junpei IKEZOE, Marc Van CAUTEREN, Joop J. van VAALS, Shinichi TAMURA, Minoru TANAKA 740-746
Key words : Interactive MR imaging, "Local Look" sequence, Interventional MR imaging
Abstract : We describe an interactive magnetic resonance (MR) imaging technique in which the scanning parameters are changed by operator interaction during continuous scanning. This technique, combined with a single-shot turbo spin-echo zoom imaging sequence, allows the operator to modify scan parameters and observe the effects on the image within a few hundred milliseconds. Specifically, the ability to perform interactive slice positioning has been demonstrated with a normal volunteer. Providing interactive scanning capabilities on a standard clinical scanner without hardware modifications offers a practical and economical method for supporting interventional and cardiac MR applications.





November(Vol.18, No.6)

[-> return]
titleauthorpage
Sensible Human Project -Towards Digital Human from Visible Human- Takashi TAKAHASHI 765-768
Key words : Visible human, Virtual reality, MRE(Magneto Resonance Elastogram), Sensible human, Virtual human, Digital human
Abstract : Visible Human Data developed by National Library of Medicine in USA have provided the electronic anatomy, promoted visualization technology for 3D tissue data and contributed promotion of medical VR expected to be a new clinical tool. The medical VR requires not only this kind of topological data but also elastic data of human tissue to express palpable feeling. In this paper, the outline of Sensible Human Project that is giving elasticity data (MRE) of human soft tissue to us by using MRI technology is briefly written. The project is expected to correspond to the Visible Human Project in the US and is also expected to extend.

[-> return]
titleauthorpage
Measurement and Evaluation of Local Tissues Elastic Modulus Using MR Elastography Mikio SUGA, Kotaro MINATO, Tetsuya MATSUDA, Osamu OSHIRO, Kunihiro CHIHARA, Makoto OHTA, Sadami TSUTSUMI, Masaru KOMORI, Takashi TAKAHASHI, Jun OKAMOTO, Osamu TAKIZAWA 769-776
Key words : Magnetic resonance elastography, Elasticity, Local wavelength estimation, Multiple phase offsets, Shear modulus
Abstract : Magnetic resonance elastography (MRE) is a phase-contrast-based method that can visualize propagating strain waves in materials. The quantitative values of shear modulus can be calculated by estimating the local wavelength of the wave pattern. Low frequency mechanical motion must be used for soft tissue-like materials, because strain waves rapidly attenuate at higher frequency. Therefore, it is difficult to estimate local wavelength with high spatial resolution especially from noisy MRE. In the MRE sequence, motion-sensitizing gradient (MSG) are synchronized with the mechanical cyclic motion. MRE with multiple phase offsets can be generated with increasing delays between the MSG and mechanical excitation. In this paper, we describe a method of measuring local wavelength with high spatial resolution using MRE with multiple phase offsets. We also describe MRE sequences and a vibration system that can drive in the gantry of a magnetic resonance imaging system. To confirm the reliability of this method, a computer simulation and phantom study were performed. The shear modulus measured with various elastic objects was well consistent with the value obtained by MRE and the mechanical method. The shear moduli of various in-vitro and in-vivo tissues were also analyzed by this method.

[-> return]
titleauthorpage
Secure Method of Making the Sensible Human Data Available on the Network Shigeki KUWATA, Hirohiko NAKANO, Yasushi MATSUMURA, Hiroshi TAKEDA 777-782
Key words : Security, Client/Server System, WWW, Attack, DoS
Abstract : It is considered most feasible to employ the web system on the internet to make the sensible human data (SH data) open and to carry out the efficient utilization of the data. This research aims 1) to develop the system suitable for making the SH data available securely on the internet, and 2) to verify whether the system can prevent threats to attack it. After developing the system using the existing information technology, we verified its security with well-known softwares for scrutinizing security holes. As a result, the system proved to have a good performance enough to be used practically against the known threats. Besides, we pointed out a security problem on the data cached in a local browser which had not been noticed thus far. We considered that installing our newly developed software would improve the quality of the system in terms of security.

[-> return]
titleauthorpage
Transmission of Bio-haptic Information in Medical Applications Masaru KOMORI 783-788
Key words : Haptic application, Delay, Surgery simulation
Abstract : A guideline of haptic delays for medical applications based on bio-elasticity data like SHP (Sensible Human Project) is described. Based on a psychological experiment, user characteristics against haptic delays such as transmission latency or insensible time due to calculation are mentioned. Unless the latency is suppressed under 30 msec and/or the insensible time under 5 msec, unnatural palpation will be obvious. A palpable cardiac model constructed under the guideline is introduced. The interactive 3D model of heart is intended to be a part of a medical education material and traning system of cardiac surgery.

[-> return]
titleauthorpage
Medical Application of Sensible Human Project Data Hiroshi OYAMA 789-793
Key words : VR, CORBA, Sensible human project, Haptic, Medical system
Abstract : This paper describes that the haptic and shape data for the human body acquired in the Sensible Human Project are used for a new application that is useful for medicine education, clinical training and minimally invasive surgery using robot technology permits the physical separation of the patient and the surgeon. I might be able to use new diagnostic rules in breast cancer, such as inspection of the incision after surgery or to compare the results of self-examination, with pathology, and haptic data first. The development of systems allowing this advance largely depends on the construction of a tactile sense presentation device. I constructed a neurosurgical training system for the National Cancer Center that constructs scenes using OpenGL Volumizer, controlling the system using CORBA technology. Research into problems created by the separation of these two is now necessary and further technical developments are required. In conclusion, haptic data on the human body may also be used in research on development of medical system.

[-> return]
titleauthorpage
Advanced Computer-aided Detection System for Microcalcifications on Mammograms by Improvement of the Morphological Filter Yohshihiro HAGIHARA, Hidefumi KOBATAKE, Shigeru NAWANO, Hideya TAKEO 795-804
Key words : Computer-aided detection, Mathematical morphology, Mammography, Micro-calcifications, Breast cancer
Abstract : This paper describes an accurate computer-aided detection system for microcalcifications on mammograms by improvement of the morphological filter. The basic system is based on morphological processing using multiple structuring elements. To improve the basic system, two new processing methods are added. The first is a histogram-based morphological opening to suppress artifacts. The other is a region expansion method that takes into consideration the gradient of the background. To evaluate the effectiveness of the proposed system, experiments on detection were performed. Test materials for the experiments were 124 CR images. The true-positive detection rate of the proposed system was 94.9% and the average number of false positives per image was 0.516. The number of false positives per image was improved by 15.8% in comparison with the conventional system. These experimental results demonstrate the effectiveness of the proposed system.

[-> return]
titleauthorpage
Basic Investigation of Fluorescent X-ray Computed Tomography Using Synchrotron Radiation for Biomedical Use Quanwen YU, Tohoru TAKEDA, Toru YASHIRO, Tetsuya YUASA, Yasuhiro HASEGAWA, Tsutomu ZENIYA, Kazuyuki HYODO, Jin WU, F. Avraham DILMANIAN, Takao AKATSUKA, Yuji ITAI 805-816
Key words : Fluorescent X-ray, Computed tomography, Synchrotron radiation, Specific materials distribution, Nondestructive measurement
Abstract : Fluorescent X-ray computed tomography with synchrotron radiation (FXCT) is being developed for imaging small amounts of target material in biomedical objects. For optimal data acquisition by a highly purified germanium detector, a performance study was carried out. Fundamental examination of FXCT was performed to assess the quantitative detectability of target material by obtaining phantom and FXCT images of the human thyroid gland. To keep the counts acquisition error below 10%, the incoming counts to the HPGe detector must be less than 10000 cps. FXCT clearly imaged 25-¦Ìg/ml iodine solution in a phantom at a 200¡ß200-¦Ìm2 spatial resolution with a 200-¦Ìm slice thickness, and a linear relation between the concentration of iodine and fluorescent X-ray counts was observed from 25¦Ìg/ml to 1000¦Ìg/ml. The two-dimensional distribution of iodine within the thyroid gland was depicted quantitatively. FXCT is thought to be a powerful tool for evaluating the distribution of a specific material in a biomedical sample.

[-> return]
titleauthorpage
Image Reconstruction by NMR Fresnel Diffractive Imaging Technique Satoshi ITO, Akiyoshi ONO, Yoshitsugu KAMIMURA, Yoshifumi YAMADA 817-827
Key words : MRI, Diffractive imaging, Fast imaging, Angiography
Abstract : A new approach to MR angiography, the NMR diffractive imaging technique, has been investigated. The expression for NMR signals obtained in the NMR diffractive imaging technique is similar to the equation for Fresnel diffraction in light waves or sound waves. Therefore, it is possible to reconstruct an image focusing on any plane in the depth direction from data scanned two-dimensionally by changing an imaging parameter in the reconstruction step. To support this imaging technique, a coil system composed of six coils was designed. Experiments were performed using an ultra-low- field MRI scanner to acquire two-dimensional data in the proposed technique. Even though blurred images outside the focus are superimposed on the image in the focal plane, the three-dimensional distribution of the object can be recognized by moving the focal plane in the depth direction. To obtain supplemental information for the object, acquiring images from different angles is helpful for recognizing the spatial distribution of the object more precisely. Although the image obtained contains blurred images outside the focus the proposed imaging technique is expected to be useful in MR fast angiography.

[-> return]
titleauthorpage
Noise Analysis and Noise Suppression with the Wavelet Transform for Low Contrast Urinary Sediment Images Ning-Feng ZENG, Keiji TANIGUCHI, Sadakazu WATANABE, Yutaka NAKANO 829-840
Key words : Noise suppression, Low contrast image, Wavelet transform, Local de-noising factor, Urinary sediment image
Abstract : This paper proposes a new scheme for noise reduction in low contrast urinary sediment images (LCUSI). The noise in LCUSI is analyzed and classified as different types of noise according to their features and source. First, a noise sample image and the features of the noise are used to reduce the non-stochastic noise. Second, the wavelet transform (a-trous algorithm) is used to divide the image into different scales for suppressing the stochastic noise separately. Then, in order to maintain the sharpness of the edges of objects, the thresholds in the different scales for filtering noise are selected by incorporating a local denoising factor which is determined by fuzzy reasoning. Finally, all the processed scales are combined. This scheme can not only effectively reduce noise in LCUSI, but also retains the information of the edges of objects in the image. Our experimental results show that the performance of this method is superior to that of other methods. This method is very effective for preprocessing LCUSI.