Abstracts of Medical Imaging Technology of 2002.

January(Vol.20, No.1) [-> return] title author page Segmentation of Medical Images Using Deformable Models : A survey Akinobu SHIMIZU 3-12 Key words : Deformable model, Snakes, Segmentation, Contour extraction, Medical image processing Abstract : This paper surveys segmentation algorithms using deformable models, which are capable of extracting objects in noisy images by using gray value features together with a priori knowledge about objects. First, modeling methods and deformation algorithms are reviewed. Then comparison between algorithms are provided concerning utilization of a priori knowledge, interaction mechanism, and extensibility to three or four dimensional images. [-> return] title author page Methods for Structural Analysis of Digital Figures Using Distance Transformation Yasushi HIRANO, Jun-ichiro TORIWAKI 13-22 Key words : Distance transformation, Thinning, Skeletonization Abstract : In this survey article, we introduce fundamental methods of the structural analysis for two-dimensional (2D) and three-dimensional (3D) figures. These methods are the basis of the procedures to extract figures or objects from an input image and to obtain shape features which characterize the extracted figures using computer. Complex image processing procedures are often composed of these basic methods. Therefore, the knowledge of the characteristics of such basic methods is useful to construct the procedures to analyze figures or to extract figures from an input image. This article presents the distance transformation, the axis thinning, the surface thinning and the skeletonization in the 2D and the 3D space with related processings including fusion, Voronoi diagram and morphological operations. A few examples of practical applications are also introduced. [-> return] title author page Image Registration Methods in Computer-Aided Diagnosis Hiroyuki YOSHIDA 23-28 Key words : Image registration, Image segmentation, Computer-aided diagnosis Abstract : Image registration has a wide utility in medical image analysis. Application to computer-aided diagnosis ranges from increasing the conspicuity of interval changes in temporal series of images to superimposing features obtained from images acquired by different modalities. This paper presents a brief overview of major registration methods based on the type of transformations and similarity measures in the registration process. A recently developed registration method that yields simultaneous segmentation and registration is described with an application to the extraction of abnormal structures that are overlapped by normal anatomic structures in chest radiographs. [-> return] title author page ROC Analysis and Evaluation of CAD System Shigehiko KATSURAGAWA 29-38 Key words : ROC analysis, Observer performance study, Computer-aided diagnosis Abstract : In order to show a CAD system to be useful for radiologists' diagnosis of medical images, it must be proved that the diagnostic accuracy of radiologists with a CAD system is superior to that without a CAD system. Sometimes we have to investigate the effect of a modification of a computerized scheme on the CAD system performance, or to compare our own CAD system with other CAD system. The ROC analysis is a powerful method for evaluation of such CAD system performance. In this article, the basic principle of the ROC analysis, and practical applications of the ROC analysis for the evaluation of the performance of CAD systems are explained. In addition, LROC and FROC analyses that are taking consideration into location of lesions are described briefly. [-> return] title author page Independent Component Analysis for Medical Image Processing Shigeru MURAKI 39-45 Key words : Independent component analysis, fMRI, Multi-channel MRI Abstract : Independent Component Analysis (ICA) can separate mixed signals without using any knowledge of the mixing process. Although it looks like magic, it is performed as a simple mathematical procedure using a very simple criterion, statistical independence. In this article, the author explains mathematical definitions of independence and shows some ICA algorithms by using our medical image processing examples, e.g. functional MRI, multi-channel MRI. [-> return] title author page Evaluation of Electrical Activity in the Human Brain Using a Dipole-distribution Model Ryoji MIZUNO, Tsuyoshi SHIINA, Yoichi SAITO 46-58 Key words : AEP, electric current dipole, dipole distribution model, SVD, generalized matrix Abstract : Evaluation of the electric current dipole distribution along the gyri is expected to enable us to assess the general activity of evoked potential in the cerebrum and to perform analysis based on functional mapping. However, most comparable current dipole analysis methods estimate and trace the positions of the sources of activity in a brain equivalent using a single electric current dipole approximation. Therefore, in the case that there is actually more than one source of activity on the cerebral surface, the results of such single dipole analysis provide an estimated position which is deeper in the cerebrum than the actual position. In this paper, we introduce a model in which many dipoles are distributed on a spherical surface corresponding to the cerebral cortex, based on a priori knowledge that the potential of pyramidal cells is higher than that of other neurons in the cerebral cortex. Although estimation of the dipole moment distribution is based on the equivalent electric current dipole method, various methods (i.e., SVD and the virtual electrode method, which we have proposed to improve the precision of estimation in the case of a small number of electrodes and a low S/N) are compared. Finally, an example of analysis using AEP when the subject is given an odd-ball task is presented and the usefulness of this proposed model is evaluated. [-> return] title author page A Parallel Registration Algorithm for DSA based on a Local Shifting Method in a PC Cluster System Raiko FUNAKAMI, Kyoichi HIROSHIMA, Junji NISHINO, Tomohiro ODAKA, Hisakazu OGURA, Testuya FUKUSHIMA, Yasuhiro NISHIMOTO, Masato TANAKA, Harumi ITO, Kazutaka YAMAMOTO 59-68 Key words : Digital subtraction angiography (DSA), Registration, Motion artifact, Parallel processing, Cluster system Abstract : This paper describes a high-speed registration method for digital subtraction angiography (DSA) in a PC cluster system connected via Ethernet. Automatic registration methods for DSA were investigated, and a registration method employing local shifting is proposed. The concurrency in this proposed registration method was examined, and several parallel registration methods were implemented in the PC cluster system. These methods were applied to DSA images. The results of the experiments showed that high-speed registration was achieved by applying the proposed parallel processing method.

March(Vol.20, No.2) [-> return] title author page Color Science and Engineering in Medicine -- Overview -- Yoichi MIYAKE 97-93 Key words : Color reproduction, Spectral imaging, Skin and membrane color, Color adaptation Abstract : Color is one of the most important information for diagnosis of various kinds of skin or membrane diseases. However, tri-stimulus values XYZ and RGB values obtained by CCD camera or color film are dependent on the spectral characteristics of imaging devices and illuminant conditions. In this paper, some problems to apply color information to diagnose are presented and our recent research on the medical color and engineering is introduced. [-> return] title author page Medical Vision Norimichi TSUMURA 94-104 Key words : Multi-spectral image, Oxygen saturation, Photometric stereo, Image based modeling and rendering, BRDF Abstract : Ubiquitous information technologies are expected to support elder people to improve their quality of life by providing medical information and managing their health information. Sensing the health information will be a key technology to achieve ubiquitous information society for elder people. In this paper, the spectral imaging technique is introduced for monitoring pigmentations of human skin by inverse optical scattering method. Photometric stereo technique is used to get the absolute spectral reflectance reducing the effects of shading and inhomogeneous illumination. The images under arbitrary illumination and point of view are reproduced using the computer graphics technique. [-> return] title author page Medical Application of Natural Vision Masahiro YAMAGUCHI 105-110 Key words : Multispectral images, Natural color reproduction, Computer aided diagnosis Abstract : "Natural Vision" is currently under development for the purpose of super-reality image telecommunication system with natural color reproduction. In this article, the medical applications of natural vision are introduced: a natural color reproduction system for telemedicine and a diagnosis support system using the multispectral images. [-> return] title author page Practical Methods of Color Quality Assurance for Telemedicine Systems Kevin M. MCNEILL, Janet MAJOR, Hans ROEHRIG, Elizabeth KRUPINSKI 111-116 Key words : Telemedicine, Teledermatology, Color quality assurance, CRT calibration Abstract : Telemedicine is experiencing significant growth in the U.S. and around the world, with a strong focus on improving the delivery of healthcare to remote rural areas. The Arizona Telemedicine Program has operated a statewide program since 1996, which provides multi-specialty care and educational content. Three of the most active specialties are image based: radiology, dermatology and pathology. Teleradiology is well established in the U.S. and well integrated into standard department practice. Teledermatology offers similar characteristics and fits well into the paradigm of remote practice, but introduces the additional factor of color fidelity. Color provides significant information for the dermatologist. It is critical that the color in a dermatology image captured at a remote site is reproduced faithfully at the consulting site. However, in many telemedicine applications the remote rural sites present significant challenges of lack of high-speed communications infrastructure, reliability of power and availability of skilled technical personnel. In this type of setting it is useful to establish practical methods for quality assurance that are inexpensive and do not rely on complex technology. The Arizona Telemedicine Program employs these practical methods in two programs involving very rural sites and has found them to be effective and appropriate for remote sites with limited resources. [-> return] title author page Significance of Color Imaging in Medical Practice -- From the Viewpoint of Medical Science -- Masahiro NISHIBORI 117-122 Key words : Medical color imaging, Multispectral imaging, Color reproduction, Diagnostic accuracy, Morphological diagnosis Abstract : Morphological diagnosis which requires reliable color imaging plays one of essential roles in medicine. However, a large potential risk of erroneous diagnoses caused by inaccurately reproduced colors has been left in digital imaging that is rapidly spreading in medicine. Aiming at ensuring enough accuracy of color reproduction for reliable diagnosis, several calibration techniques for colorimetric color reproduction may be put into practice, yet they have serious weak points. Multispectral imaging is the most hopeful technology that would provide perfect solution to medical color complications when it will be available. Until the time comes, every endeavor to ensure the 'diagnostic equivalence' among various display equipment should be required. An important question is why medical practitioners who are most nervous about erroneous diagnosis do not complain of inaccurate color reproduction. Our experience in collaborative study with researchers from various medical subfields and from the color engineering field has revealed the very reason. That is, most of them have so simplified color conceptions, for example, 'The three primary colors can reproduce every color because we have only three kinds of cone cells,' that they believe present performance of color reproduction reaches a theoretical limit. Therefore, we should make them realize the truth of color by any means. [-> return] title author page Real-time Multi-spectral Image Processing for Mapping Pigmentation in Human Skin Daisuke NAKAO, Norimichi TSUMURA, Yoichi MIYAKE 123-133 Key words : Real-time image processing, Oxygen saturation, Skin pigmentation, Principal component analysis, Real-time Abstract : In the field of medical imaging, skin responses to various stresses provide important information for monitoring the subject's physiological condition. In our research, we have developed a real-time video signal processing system that can estimate and display maps of pigmentation in human skin. Maps of melanin, oxyhemoglobin, and deoxyhemoglobin concentrations in skin are generated from RGB images using three precomputed color conversion look-up tables. This system was found to be useful for observing changes in pigmentation during the occlusion and restoration of blood flow to the human middle finger and also during anaerobic exercise. It is possible to monitor changes in human physiological homeostasis in real time using this system. [-> return] title author page Recognition of Abnormal Substances in Urinary Sediment Images Using a Neuro-Fuzzy Technique Ning-Feng ZENG, Sadakazu WATANABE, Keiji TANIGUCHI, Yutaka NAKANO 134-145 Key words : Pattern recognition, Fuzzy representation, Neuro-fuzzy technique, Abnormal sub-stance, Urinary sediment images Abstract : We present a method for recognizing abnormal substances in a urinary sediment image (USI) using a neuro-fuzzy technique. This technique has the advantages of both neural networks and fuzzy inference models. With regard to the pattern observed for abnormal substances, which are abnormal in shape and size, their feature data are incomplete and the observations of these sub-stances are imperfect. Fuzzy techniques are a good way to represent them. Although fuzzy inference models are effective for dealing with the classification problem, they suffer from some limitations in precision. This paper describes the recognition of white blood cells as an example. First, the distribution of feature values is obtained from a group of sample data. Then, based on this distribution, the initial parameters of the fuzzy sets are estimated. After this, the neuro-fuzzy sys-tem is trained using a group of sample data to refine the parameters of the fuzzy sets. Finally, the trained networks are used to detect the unknown substances. Our experimental results show that the neuro-fuzzy technique is a convenient method for obtaining precise results as compared with fuzzy inference models. [-> return] title author page Archiving Period and Frequency of Use of Pictures in PACS Tsuneaki NAKAGAWA, Kazuo NIWA, Isamu OHASHI, Hitoshi SHIBUYA 146-150 Key words : PACS, Hierarchical storage management, Digital tape archive, Archiving period of medical images Abstract : The appropriate period for archiving pictures in a picture archiving and communications system (PACS) has not yet been established. To address this issue, we studied the relationship between the archiving period and the frequency of use of pictures at our institution. The result was that pictures with a longer archiving period were used less frequently, with a negative logarithmic correlation observed between the two factors. It was also found that the frequency of use decreased by one half for every 7 months of archiving at our institution. This characteristic of the frequency of use of pictures should be considered when establishing the archiving period for PACS.

May(Vol.20, No.3) [-> return] title author page Evaluation of Brain Function and Metabolism using MRI Masafumi HARADA, Sonoka HISAOKA, Mayumi TAKEUCHI, Hiromu NISHITANI 156-161 Key words : fMRI, MRS, BOLD, Flow, Metabolism Abstract : The techniques for evaluation of brain function and metabolism by MRI are introduced, and the possibility and issues are also discussed in this paper. To detect activated area of the brain by some task, the blood oxygenation level dependent (BOLD) methods is widely used in this 10 years and the several different ways for statistical analysis were reported so far. The classical method is t-test which is included in SPM, the most popular fMRI software. But non-parametric analysis or non-prior knowledge blind component methods, for example independent component analysis were developed and its usefulness was reported recently. However relationship between hemodynamic change and neuronal activity was still unclear and remained for future study. To evaluate metabolites in the brain, proton MRS is used by the clinical instrument, and the number of papres which reported its clinical usefulness is gradually increased. Recent work of proton MRS on psychiatric disorder showed a interest result as decrease of Glutamine and Glutamate concentration. The sensitivity and reproducibility should be evaluated carefully, but the techniques for evaluation of brain function and metabolism by MRI is more developing and will offer important and useful information for both neuroscience and clinical diagnosis. [-> return] title author page Clinical Applications of Near Infrared Spectrography Eiju WATANABE 162-167 Key words : Near infrared, Noninvasive functional brain mapping, Language dominancy, Epilepsy Abstract : Near infrared spectroscopic topography is a new alternative of noninvasive functional brain mapping technology. Reflection of near infrared light projected into the brain from scalp surface is utilized by this system. We assume that the reflection represents the hemoglobin concentration change which is parallel to the regional brain activity. We are using this system in detecting language dominant hemisphere, epileptic focus or motor cortex. The spatial resolution is 20mm and time resolution is 200msec. The biggest advantage of this system is that the system is very tolerant to the movement so as to be used in a condition similar to the daily life. Various further applications will be considered in the fields of brain physiology and psychology. [-> return] title author page Neuromagnetic Studies using SQUID Magnetometer Keiji IRAMINA, Shoogo UENO 168-173 Key words : SQUID, MEG, Neuromagnetism, Higher brain function, Functional imaging Abstract : It is now possible to visualize human's neural processes non-invasively by measurement of magnetoencephalography(MEG) using SQUID (Superconducting Quantum Interference Devises) magneometer. With the advance of SQUID magnetometer which is able to cover the whole head, MEG become a useful method for neuroscience study with a few mm spatial and 1 ms temporal resolution. In this paper, we reviewed recent studies of neuromagnetism, especially higher brain function such as attention, perception, and language processing. We also present the comparison between MEG and fMRI (functional magnetic resonance imaging) which recently comes into the limelight as the noninvasive method of functional brain study. [-> return] title author page Biomedical Functional Imaging using Positron CT and its Application to Clinical Research Hidehiro IIDA 174-183 Key words : Positoron CT, Kinetic analysis, Physiological function Abstract : Positron emission tomography is capable of providing various physiological functions in vivo. Extremely high sensitivity allows the imaging of organic functions of various receptors, namely the molecular imaging. This is based on quantitative imaging of the radio-tracer distribution in organs and the mathematical modeling of their kinetic behavior in the body. Information obtained by this technique is invaluable for clinical research for understanding the patho-physiology in cardiovascular diseases. [-> return] title author page X-ray Imaging with Monochromatic Synchrotron Radiation: Fluorescent and Phase-contrast Method Tohoru TAKEDA, Yuji ITAI 184-193 Key words : Synchrotron radiation, Element analysis, Vessel imaging, Fluorescent x-ray CT, Phase-contrast x-ray imaging Abstract : To obtain the high sensitive x-ray images of biomedical object, new x-ray imaging techniques using fluorescent x-ray and phase-contrast x-ray are being developed in Japan. Fluorescent x-ray CT can detect very small amounts of specific elements in the order of ppm at one pixel, whereas phase-contrast x-ray imaging with interferometer can detect minute differences of biological object. Here, our recent experimental results are presented. [-> return] title author page Improvement of Lung Cancer Detection Method Based on Ground Glass Opacities in X-ray CT Images Hotaka TAKIZAWA, Makoto IKARASHI, Toshiaki OKUMURA, Shinji YAMAMOTO, Tohru NAKAGAWA, Tohru MATSUMOTO, Yukio TATENO, Takeshi IINUMA, Mitsuomi MATSUMOTO 194-202 Key words : Chest X-ray CT images, Lung cancers, Ground glass opacities, Automatic detection, Discriminant functions Abstract : In this paper, we describe a method for the automatic detection of ground glass opacities, which are observed in early-stage lung cancers, in computer-aided diagnosis employed in lung cancer screening by CT (LSCT). First, candidate shadows detected by our Variable N-Quoit filter are classified by size and position. Then, cancer shadows are distinguished from normal shadows using discriminant functions derived from feature analysis. As a result, false-negative and false-positive rates are reduced. [-> return] title author page Coarse Calcification Detection Method and its Application to Reducing False Microcalcification Clusters Nobutada MIZUSAWA, Yoshihiro HAGIHARA, Akinobu SHIMIZU, Hidefumi KOBATAKE, Hideya TAKEO, Shigeru NAWANO 203-211 Key words : CAD system, Mammogram, Microcalcification, Coarse calcification Abstract : The presence of microcalcification clusters is useful sign for the detection of malignant tumors in the early stages, and the identification of such tumors is important in mammography CAD systems. The existence of coarse calcifications is not related to the presence of a malignant tumor, but their local irregular intensities tend to be detected as microcalcifications. This paper presents a method for extracting coarse calcifications and for reducing false microcalcification clusters on digital mammograms. The extraction process consists of two steps, rough extraction and precise extraction. Coarse calcifications can be extracted efficiently and precisely. The extracted regions are then used to reduce false clusters. Sixty mammograms were used in this study, and the results showed that 14 false microcalcification clusters in 17 clusters caused by coarse calcifications were eliminated without deleting true clusters. [-> return] title author page Iterative Signal-to-Noise Ratio Improvement Using Dual Fresnel Transform Pairs in Magnetic Resonance Imaging Satoshi ITO, Fujitaka SUZUKI, Yoshifumi YAMADA 212-216 Key words : SNR, MR image, Fresnel transform, Wiener filter, Multi-resolution analysis Abstract : This paper describes an iterative image restoration technique based on the Fresnel transform band-splitting effect (FREBAS) that can be applied to improve the signal-to-noise ratio (SNR) of magnetic resonance (MR) images. The algorithm is based on the principle of adaptive image analysis via an arbitrary scaling parameter D. Iterative SNR improvement using FREBAS and Wiener filtering is performed by calculating the Wiener function for a pre-filtered image. It is shown that image quality can be improved iteratively using this technique. Image SNR improvement is compared with other methods such as adaptive nonlinear filtering and 2-step Wiener filtering in the wavelet domain and it is demonstrated that the proposed method provides the greatest SNR improvement and that the resultant image is free of serious image degradation. The proposed denoising method is a practical filter that offers a high degree of SNR improvement while maintaining phase information.

September(Vol.20, No.5) [-> return] title author page Image Recognition and Process of the Image Diagnosis An example of image diagnosis with chest radiograph Junpei IKEZOE 535-539 Key words : Image recognition, image diagnosis, Chest radiography, Clinical information Abstract : The purpose of this short communication is to explain the image recognition and process of diagnosis made by radiologists. The points are 1) clinical information, 2) purpose and methods of examination, 3) image findings, 4) diagnosis with image findings and clinical information. [-> return] title author page Role of Medical Imaging Technology in Multislice CT Innovation Kazuhiro KATADA 540-544 Key words : X-ray computed tomography, Multislice CT, Isotropic voxel, Computer-aided diagnosis Abstract : Recent evolution of multislice CT technology enabled us to obtain true isotropic volumetric data. As a result, CT data explosion has become a serious problem in clinical field. Medical imaging technology may play important role in terms of reconstruction algorithm, data handling, noise reduction, multi-purpose CAD (computer-aided diagnosis) and area detector CT system. Intense co-operation between medical and technological staff is essential to obtain successful results. [-> return] title author page MRI Hiroshi FUKATSU 545-550 Key words : MRI, DICOM viewer, Filmless, Reporting system Abstract : MRI is one of the modern sophisticated diagnostic tools because without X-ray irradiation. At the same time, with its superior intrinsic soft tissue contrast, multi-parametric functional information, MRI should be the mainstream in 21st century in diagnostic imaging. On the other hand, because of its complicated information and vastness of the information amount, diagnostic radiologists are now expected to answer the clinical question with specific opinion as specialists. The workflow as a radiologist include the understanding the patient's history and present illness, past history and the information concerning the contraindication, purpose of the study, findings of the previous study or those of other modalities, understanding of the physics of the modality and the relationship of the imaging findings with pathophysiological condition. In addition, number of cases a day are rapidly increasing. This enormous information is now threatening the foundation of the diagnostic imaging paradigm. In this short article, we try to discuss about the desired system to make the most of the advanced computer and network system. [-> return] title author page Medical Imaging Technology: Expectations from Interventional Radiology - Three-dimensional Imaging and Radiation Protection - Tsuneo ISHIGUCHI 551-555 Key words : Aortic aneurysm, Stent graft, Multislice CT, Radiation protection, digital imaging Abstract : Interventional radiologic technique is minimally invasive to patient and now it is used for the treatment of many diseases. Precise imaging is mandatory in the planning of treatment. Three-dimensional imaging with multislice helical CT is important for vascular interventional radiology including stent-grafting for aortic aneurysms. Precise vessel measurements and simulation software of the treatment will result in improvement of the success rate and long-term results. Most interventional radiologic procedures are performed under x-ray fluoroscopic guidance. A complicated procedure needs long fluoroscopic time and many image acquisitions associated with high radiation exposure to the patient's skin, which may result in radiation skin injury including ulcerations or necrosis. Improvements in high-sensitivity detectors and image processing and display techniques are necessary for the development of a low-dose and high-quality imaging system dedicated for interventional radiology. [-> return] title author page PET/SPECT Kengo ITO 556-561 Key words : PET, SPECT, Semiconductor detector, Fusion image, Voxel based analysis Abstract : This article focuses on clinical aspects of PET/SPECT technology in the near future. In order to perform high sensitivity and high spatial resolution in PET/SPECT, new technologies related to detector such as semiconductor detector are highly expected. High sensitivity and high spatial resolution can realize both high image quality and shortening of scan time. Fusion images with multi-modality system such as CT/PET and CT/SPECT may become more popular, and the application of fusion images for clinical management may expand the usefulness of nuclear medicine imaging. For interventional radiology, an open MRI system combined with PET may be more useful. A voxel based statistical analysis of images with image database can improve the diagnostic accuracy and then contribute to clinical practice. [-> return] title author page Development of a Function for Identifying Nonvisualized Regions in a Virtual Endoscope System Yuichiro HAYASHI, Kensaku MORI, Jun-ichi HASEGAWA, Yasuhito SUENAGA, Jun-ichiro TORIWAKI 562-571 Key words : Virtual endoscope system, Virtual colonoscopy, Nonvisualized regions, Fly-through Abstract : This paper presents a function in a virtual endoscope system (VES) for recording regions that are not visualized by the user during fly-through and to present regions that are not displayed on the screen at all. In diagnosis using a VES, the physician performs fly-through within organs that are reconstructed in a computer and observes polyps by changing the viewpoint and the viewing direction. It is critically important that the entire organ be examined in order to avoid the overlooking of finding lesions. However, there is the possibility that the physician may terminate the VES examination without observing the target organ in its entirety, especially when the organ has a complex anatomy. In this paper, we describe a newly developed function that detects regions that were not visualized during fly-through. This function permits the user to identify nonvisualized regions and draws the user's attention to these regions. The proposed methods for detecting nonvisualized regions is applicable to both the surface-rendering and volume-rendering methods employed in VES examinations. In the surface-rendering method, we record triangular patches that were not displayed on the screen. In the volume-rendering method, the system identifies the voxels that were not displayed on the screen by analyzing the changes in summed opacity values along the rays generated for rendering. The proposed functions were implemented in our VES. In order to evaluate the performance of the proposed methods, we applied them to the evaluation of an artificial object, the colon, and the bronchial region. The experimental results showed that these methods were effective in identifying nonvisualized regions. [-> return] title author page Recognition of the Aorta and Pulmonary Artery in the Mediastinum Using Medial-Line Models From 3D CT Images Without Contrast Material Takayuki KITASAKA, Kensaku MORI, Jun-ichi HASEGAWA, Jun-ichiro TORIWAKI, Kazuhiro KATADA 572-583 Key words : 3D chest X-ray CT image, Aorta, Pulmonary artery, Medial line model, Euclidean distance transformation Abstract : In this paper, we propose a new method for automated extraction of the aorta and pulmonary artery (PA) regions in the mediastinum from non-contrast 3D chest X-ray CT images using medial line models. The proposed method does not extract the contours of these blood vessels directly, but extracts the medial line of each vessel first and then recovers each vessel area. First, the process performs edge detection based on the local standard deviation to obtain the edge areas of the vessels. Second, the Euclidean distance transformation is applied for non-edge areas, and likelihood images of the centers of the vessels are obtained. The medial-line models are deformed based on the likelihood images so as to fit them to the center of each artery. The aorta and pulmonary artery regions are obtained by employing reverse distance transformation to the medial lines extracted above. We applied our method to noncontrast 3D chest X-ray CT images in seven cases and compared the extraction results obtained by this method with the results obtained by manual input. The experimental results for the extraction of the aorta and the PA showed coincidence indices of approximately 90% and 80%-90%, respectively, confirming that the aorta and PA regions could be extracted satisfactorily. [-> return] title author page Tractography of Brain White Matter Tracts Based on Analysis of MR Diffusion Tensor Data: Selective Tractography with Tracking Reliability Yoshitaka MASUTANI, Osamu ABE, Shigeki AOKI, Harushi MORI, Tomohiko MASUMOTO, Takeharu YOSHIKAWA, Kuni OHTOMO, Naoto HAYASHI, Hiroyuki KABASAWA 584-592 Key words : MR diffusion tensor, Diffusion anisotropy, Visualization, Tractography Abstract : A new display method for tractography in magnetic resonance diffusion tensor imaging has been developed. In the conventional methods for tractography that have previously been reported, only paths of maximum diffusion tracking are displayed. However, due to low anisotropy, the direction of maximum diffusion does not always correspond to the direction of tracts. In the proposed method, the local reliability of diffusion tracking is evaluated based on anisotropy, and paths of tracking are displayed using gradation of colors and degrees of opacity according to the tracking reliability. The method was validated by using several clinical data sets obtained from volunteers and patients. [-> return] title author page Integrated Visualization System for MEG and MRI Data Akira UCHIDA, Hiroshi ENDO, Tsunehiro TAKEDA 593-598 Key words : MEG, MRI, 3-D visualization, Brain function Abstract : In MEG data analysis, we estimate current sources from the measured magnetic field to identify activated areas in the cerebral cortex. Since such estimates have many possible solutions, we need to judge the validity of the results with the aid of an anatomical brain structure and MEG data patterns simultaneously. Therefore, systems for visualizing both the measured magnetic fields and the estimated current sources with three-dimensional anatomical structures are indispensable for the analysis. We have developed such a system, in which MEG and MRI data are processed systematically, using AVSTM visualization software (Advanced Visual Systems Inc., USA). This system can show the magnetic field patterns simultaneously in both two-and three-dimensional representations. At the same time, the system provides the end user with a high degree of flexibility due to the characteristics of AVSTM. As a result, anyone can add new processing functions to the analysis system. In this paper, we describe the system that we have developed and discuss its advantages in analyzing actual MEG data.

November(Vol.20, No.6) [-> return] title author page Blood Flow Simulation in Cardiovascular System Hao LIU, Takami YAMAGUCHI, Ryutaro HIMENO 615-621 Key words : Predictive medicine, Cardiovascular blood vessel, Medical image, Hemodynamics, Computational biomechanics Abstract : Computational modeling of haemodynamics in cardiovascular system can be a useful means for predictive medicine, which may be implemented in a form of Computation-Aided Diagnosis (CAD) and/or Computation-Aided Surgery (CAS). A basis for such a new paradigm, in particular the CAD and CAS systems, for the coherent complexity in morphology and physiology of circulation system, requires an efficient and robust, PAtient-Specific Simulator (PASS) for haemodynamics. In this paper we highlight those issues relating to the establishment of such a PASS system, which is under development in RIKEN and present some preliminary results. [-> return] title author page Computational Fluid Dynamics for Cerebral Aneurysms Kazuaki FUKASAKU, Makoto NEGORO 622-630 Key words : Cerebral aneurysm, Computational fluid dynamics, Subarachnoid haemorrhage, Embolization, Indication Abstract : Cerebral aneurysms are common (1กม9 % of population) disease and the commonest cause of subarachnoid haemorrhage, which has poor prognosis even recent development of medical techniques. On the other hand, the course of unruptured aneurysms is not clear, at present. Several clinical studies to follow such unruptured aneurysms for years have been started. The result may be evidential as mass study, but not always effective to precise risk of rupture for each aneurysm. Intra-aneurysmal pressure may be one of the possible cause of rupture and affected by systemic blood pressure and intracranial pressure. Shear stress affects endothelial cell to produce vascular remodelling which has possibility to produce aneurysm and/or aneurysm wall. Both of intra-aneurymal pressure and wall shear stress are affected by flow. Here, computational fluid dynamics has great possibility to predict risk of rupture for each aneurysm. Moreover, CFD can support intravascular surgery for aneurysms when detailed three dimensional structure of coils, vessels and aneurysms will be able to obtain while the procedure. [-> return] title author page Eyeball Mechanical Simulation in RIKEN Hideo YOKOTA, Akitake MAKINOUCHI, Zhi-Gang SUN,Hiroo YABE 631-638 Key words : Eyeball, Retinal detachment,Finite element analysis, Nonlinear large deformation, Biomechanical simulation Abstract : The aim of our research project is to develop a biomechanical simulation system of the soft tissues and organs, especially the eyeball. In order to achieve this end, the system will consists of the following issues. 1) Collection of eyeball shape information, 2) Segmentation of eyeball tissues using full-color image obtained in 1), 3) measurement of mechanical properties of eyeball soft tissues, 4)mesh generation for finite element analysis, 5)development of a 3-D FEM program for coupling analysis of hyperelastic solid and static liquid. These issues are now under development and a simulator of eyeball encircling buckling operation,which is frequently applied to treatment of the eyeball retinal detachment in clinic, is expected to be carried out in March 2004. [-> return] title author page Computer Simulation for Several Ophthalmic Diseases Hiroo YABE, Ryuhei KAWAGUGHI, Zhi-Gang SUN, Hideo YOKOTA, Akitake MAKINOUCHI 639-646 Key words : Computer simulation, Ophthalmic disease, FEM, Orbital fracture, Blunt ocular trauma, Retinal detachment, Glaucoma Abstract : The mechanisms for several ophthalmic diseases in respect of their etiology and surgical procedures are amenable to the laws of physics. We studied the clinical application of our simple model of the eye and correlated the results of this simulation with the clinical events. Individual patient data could in future be input to an accurate simulation model of the eye so that we can visually understand the mechanism for a disease and apply new treatments and surgical procedures without applying undue physical and psychological load on a patient. This paper reports the progress in our studies on computer simulation for several ophthalmic diseases. [-> return] title author page Three-dimensional Remodeling Simulation for Trabecular Structural Adaptation by Using Image-based Models Ken-ichi TSUBOTA, Taiji ADACHI, Yoshihiro TOMITA, Akitake MAKINOUCHI 647-653 Key words : Computational biomechanics, Bone remodeling, Adaptation, Cancellous bone, Image-based model Abstract : Bone functionally adapts its structure to the external-loading condition due to remodeling that is driven by local mechanical stimuli. A computational mechanics approach with detailed morphological modeling of bone structure is effective understanding the mechanism of adaptive bone remodeling. In addition, computational remodeling simulation is widely expected to become a useful tool in medical and engineering fields. in this paper, we focus on the structural element of cancellous bone, so called trabecula, and illustrate a three-dimensional remodeling simulation for bone structural adaptation using image-based models. The simulation method is useful for predicting trabecular bone remodeling phenomenon in individual patients because of its applicability to the medical image data, as well as for developing a computational design system for bone-implant such as a femoral stem of artificial hip joint. [-> return] title author page Anatomical Aspects of Airway Flow and Gas Transport in the Pulmonary System Toshihiro SERA, Kazuo TANISHITA 654-659 Key words : Central-Airway, Small Airways, Alveoli Abstract : The studies of respiratory mechanics have been advanced over the past several years. They dealt with normal physiologic function, environmental health science and therapy involving a wide variety of mechanical, chemical and biomedical engineering. The airway anatomically is comprised of a branching network of numerous tubes, and so these respiratory mechanics is directly related to it's complicated structure. The subjects of respiratory mechanics involve the nasal flow, aerosol transport, airway stability, pulmonary acoustics, and surfactant dynamics. In this review, we mainly focused on anatomical aspects of "respiratory flow and transport in the pulmonary system", A) Central-airways, B) Small airways, C) Alveoli. [-> return] title author page Destructed Imaging of Biological Sample Using a 3 Dimention Internal Structure Microscope Hideo YOKOTA, Sakiko NAKAMURA, Ryuhei KAWAGUCHI, Akitake MAKINOUCHI, Hiroo YABE, Toshiro HIGUCHI 660-665 Key words : Slice, Cross sectional image, Reconstructed 3D image, Full color, Biological sample Abstract : The demand of the observation of the 3 dimensional structure has risen intended for the biological sample. In these observation methods, the nondestructive observation method of X-ray CT and MRI etc. are a main current. However, the destruction observation method has high resolution and the detectability compared with a nondestructive observation. In this paper, the 3 dimention internal structure microscope observed while automatically destroying the 3 dimensional structure of the biological sample is described. [-> return] title author page Various Functions of MRI System - From Shape to Elasticity - Yoko KATO, Ryutaro HIMENO 666-670 Key words : MRI, Relaxation time, Flow velocity, Elasticity Abstract : MRI is the system to get the information about the inside of the body non-invasively. But it is also remarkable characteristics that MRI provides various kinds of information. The signal could be evaluated quantitatively because there are the functions to express the signals. We introduce the necessary parameters for the computational biomechanics and the sequences to get the parameters. [-> return] title author page Role of Computer-Assisted Diagnosis in Mass Screening for Cancer Takeshi IINUMA, Tohru MATSUMOTO 671-675 Key words : Computer-assisted diagnosis, Cancer screening, Mammography, Lung screening CT, Dual reading Abstract : Computer-assisted diagnosis (CAD) systems are expected to play an important role as a secondary reader in screening examinations in cancer mass screening programs. In Japan, mammography is employed as a screening examination for breast cancer, and lung screening CT (LSCT) is currently under investigation as a potential replacement for conventional chest X-ray in lung cancer screening. In such screening examinations, dual image reading by two independent physicians is recommended. A CAD system may be able to replace one of these physicians, and the other physician should read all images with the assistance of the CAD system in order to determine the need for detailed examination. In order to achieve this goal, the CAD system must be evaluated with the cooperation of the reading physician and the results obtained must be compared against those of dual reading by two physicians employing the same images. If the CAD+physician method is found to provide better results than dual reading, the former should be adopted in place of the latter. This paper presents a detailed discussion of the methods that should be employed in such reading experiments. [-> return] title author page An Intuitive Display System for Observing Movement of the Diaphragm in the Diagnosis of Lung Diseases Yoshio YANAGIHARA, Hiromitsu HAMA, Mitsuru KOMIZU 676-684 Key words : Diaphragm movement, 3D magnetic sensor, Intuitive control, Observation system Abstract : Observation of the localized movements of various parts of the human diaphragm provides important information in the diagnosis of lung diseases. However, it is not easy to observe all areas of the diaphragm during breathing in separate tomographic images. For the qualitative observation of the movements of the diaphragm, we have therefore developed a new computerized system to generate dynamic volume images from sagittal and frontal MRI images of the lung and to allow the optimal direction and position for observation of these dynamic volume images to be selected in an easy and intuitive manner. Our experimental results showed that the area of the extracted region of the diaphragm and thoracic wall was almost equal (with a mean difference of approximately 0.9%) to that obtained by manual identification of the same region. The response time for redrawing a volume image from another view is less than about 32.8 ms, which indicates that the system should be suitable for practical application with intuitive control. We will also attempt to develop a method for measuring the movements of the diaphragmusing this system for CAD. [-> return] title author page Extraction of the Glomerular Region in Kidney Images Using a Local and an Elliptical Global Method Keiko HIRAI, Sadakazu WATANABE, Keiji TANIGUCHI, Yutaka NAKANO, Hong ZHU 685-693 Key words : Renal glomerular region, Bowman's capsule, Boundary estimation, Segmentation model Abstract : This paper deals with a local and a global method for segmenting the glomerular region in kidney images. In this technique, an image of a sliced and stained sample of the kidney is acquired using a CCD camera connected to an optical scanning microscope. The original image of the glomerular region is binarized by a local threshold method. The external margin of Bowman's capsule is extracted as a segment of the border region. If the extracted external margin is not closed, the border of Bowman's capsule is calculated iteratively until the segment of the external margin in the extracted border region is fitted to either a circle or an elliptical curve. After the above iteration, the curve showing the best fitting is selected as the border of Bowman's capsule. The experimental results showed that this method is more effective than conventional methods for the segmentation of the glomerular region in kidney images. [-> return] title author page Analyzing the 3D Shape and Respiratory Motion of the Rib Using Chest X-Ray Images Myint Myint SEIN, Mitsuru KOMIZU, Yoshio YANAGIHARA, Hiromitsu HAMA 694-702 Key words : Respiratory motion of the ribs, B-spline curve control points, Measurement matrix, Singular value decomposition, Thorax, Chest X-ray images Abstract : This paper presents a new method for analyzing the 3D shape and respiratory motion of the ribs using 2D X-ray images. Each rib is represented as a curve, and its control points are estimated by a cubic B-spline curve iteration algorithm. In order to recover the 3D motion and the control points of the ribs from these 2D control points, a factor decomposition algorithm for projective structure and motion has been developed. A measurement matrix is generated based on the corresponding control points of each curve. By using these control points rather than all points on the curve, the data capacity and computational time are reduced. Furthermore, the precision of matching pairs of ribs in multiple images is improved. Estimation errors arise not only from the 2D control point approximation but also from the computation of the 3D control points. Experiments with X-ray images have been conducted and the effectiveness of the proposed technique has been confirmed by obtaining results with acceptable errors. This approach may prove useful for the evaluation of chest expansion in the assessment of pulmonary diseases.