[en] Many types of monochromatic X-ray CT with synchrotron radiation (SR-CT) is being developed in the biomedical fields because of its' sufficient X-ray flux and tunability of energy spectrum. We introduce a SR-CT system to reveal the fine structures of small animal, and to detect a tracer material for functional evaluation of the organs. In the high spatial resolution SR-CT, the detailed structures of skull of live rat was demonstrated 36-μm spatial resolution with 36-μ-m slice thickness. In the high contrast SR-CT, phantom examination with contrast material (iodine) was performed. The 448 μg/ml iodine contrast material was identified. Image quality was not sufficient because of insufficient dynamic range of our sensor 'IP', insufficient dynamic range of our sensor 'IP', small number of projection, scatter contamination and 99 keV contamination (8%). Remarkable improvement is expected by using solid state detector, increasing scan projection, decrease the scatter reduction, and so on. (author)

[en] Accurate reconstruction of the pelvic structure is the most important factor to obtain desirable results after anorectal surgery. Preoperative evaluation of the anatomy is indispensable for choosing an appropriate operative method in each case. To facilitate the preoperative evaluation, three dimensional images of the pelvic structure of patients with anorectal malformations were constructed by computer graphics based upon two dimensional images obtained from MR-CT. Graphic data from MR images were transferred to a graphic work station. The anorectum, bladder and sphincter musculature were displayed three-dimensionally after segmenting these organs by (1) manually regioning the area containing the specific organ and (2) thresholding the area by the T₁ intensity level. The anatomy of each type of anomaly is easily recognized by the 3-D visualization of pelvic viscera and sphincter musculature with emphasis on position and shape of the musculature although there are some difficulties to visualize soft tissue organs. The advanced programs could show the graphic images from any desirable angle quickly enough to be helpful for the simulation of the surgery. Three-dimensional display can be very useful for better understanding of each anomaly and determining the operative method prior to surgery. (author)

[en] A method for reducing the number of X-ray CT images (30 images per patient) to be presented to doctors, is developed. This method is primarily based on MIP (Maximum Intensity Projection) processing. And a precise extraction algorithm of lung area for each slice is newly developed, which is applied before MIP processing. This method is examined by 7 samples of lung cancers, and is proved to be very effective and practical. (author)

[en] This paper describes a new system in which a user can directly watch and manipulate a true 3-D medical image, as if he/she actually watches and manipulates the true object. These characteristics were achieved by a volume scanning display for presenting a true 3-D image directly from volume data, and the Boolean set operations on the volume data for realizing real-time image manipulation. The simple example using X-CT head data has also been presented. (author)

[en] Many algebraic reconstruction techniques (ART) image reconstruction algorithms, for instance, simultaneous iterative reconstruction technique (SIRT), the relaxation method and multiplicative ART (MART), have been proposed and their convergent properties have been studied. SIRT and the underrelaxed relaxation method converge to the least-squares solution, but the convergent speeds are very slow. The Kaczmarz method converges very quickly, but the reconstructed images contain a lot of noise. The comparative studies between these algorithms have been done by Gilbert and others, but are not adequate. In this paper, we (1) propose a new method which is a modified Kaczmarz method and prove its convergence property, (2) study performance of 7 algorithms including the one proposed here by computer simulation for 3 kinds of typical phantoms. The method proposed here does not give the least-square solution, but the root mean square errors of its reconstructed images decrease very quickly after few interations. The result shows that the method proposed here gives a better reconstructed image. (author)

[en] This study was undertaken to compare two kinds of analytical attenuation compensation methods: Bellini's method using average conjugate projection views, Inouye's method. Simulation and phantom studies were performed to evaluate a linear correlation of radioactivities and reconstructed values using a water-filled cylinder phantom of 20 cm in diameter containing seven hot rods of 3 cm in diameter of different activities. Both of Bellini's and Inouye's methods showed the same high coefficient of linear correlation (R² =0.996). Therefore, the fact showed that both methods provided the accurate reconstruction. We also evaluated the statistical noise using the water-filled cylinder phantom of 20 cm in diameter and elliptical cylinder phantom with dimensions 18 cm by 14 cm, and found that the percent root mean squares errors (%RMSE) in Bellini's method were smaller than those in Inouye's method. Significant linear correlation was observed in the reconstructed values by two methods on a brain phantom (R² =0.978) and ^99mTc-ECD brain perfusion studies by a normal volunteer (R²=0.987). (author)

[en] Image registration of functional (SPECT) and morphological (X-ray CT/MRI) images is studied in order to improve the accuracy and the quantity of the image diagnosis. We have developed a new frameless registration method of X-ray CT and SPECT image using transmission CT image acquired for absorption correction of SPECT images. This is the automated registration method and calculates the transformation matrix between the two coordinate systems of image data by the optimization method. This registration method is based on the similar physical property of X-ray CT and transmission CT image. The three-dimensional overlap of the bone region is used for image matching. We verified by a phantom test that it can provide a good result of within two millimeters error. We also evaluated visually the accuracy of the registration method by the application study of SPECT, X-ray CT, and transmission CT head images. This method can be carried out accurately without any frames. We expect this registration method becomes an efficient tool to improve image diagnosis and medical treatment. (author)

[en] The author's three surgical simulation systems are presented. First the computer graphics surgical simulation system has been developed which make the three dimensional skull image from CT scans and the arbitrary osteotomy, mobilization of bone segments and prediction of post-operative appearance is made possible. The second system is solid modeling of the skull using laser curable resin and it is concluded that life-sized skull model is useful not only for surgical simulation of major craniofacial surgery but also educational purposes. The third one is solid modeling of the ear using non-contact 3-D shape measurement with slit laser scanner. The mirror image life-sized wax model is made from the normal side of th ear and the autologous cartilage framework is assembled to simulate the wax model, thus the precise three dimensional reconstruction of the auricle is made possible. (author)

[en] Synchrotron radiation (SR) has the following excellent properties as an X-ray source: high brilliance, broad continuous spectrum and small divergence These properties are advantageous for the computed tomography to require high quantitative analyses, since a high flux of monochromatic synchrotron X-ray brings an approximately twofold increase in contrast of tracer material, e.g., iodine, compared to a conventional X-ray obtained from an X-ray tube because of no beam hardening, and its small divergence can yield a considerable reduction of scattered X-ray dose owing to the wide air gap distance. Furthermore, the K-edge energy subtraction method, where subtraction is performed between CT images obtained at the energies above and below the K-absorption edge of target material, may reveal only the spatial distribution of target material with high spatial and contrast resolutions (a few mm for a few 10 μg/ml iodine solution). Therefore, we constructed a prototype CT system for biomedical use using Imaging Plate as a detector at the beamline of NE-5A in Tristan Accumulation Ring, KEK, and performed preliminary experiments to confirm the effectiveness of SR-CT. From experimental results, the detectability of iodine in our system was approximately 500 μg/ml. In this paper, the detectability of tracer material in above system is discussed theoretically and experimentally. Consequently, in order to detect a target concentration (a few 10 μg/ml), we must use the detector which is capable of employing a high-resolution AD converter (18-20 bits) and has high sensitivity in a high flux area (10⁷-10⁸ photons/mm²/s). The realization of a SR-CT system would permit quantitative functional evaluations for the regional blood flow and the metabolism in organs with high accuracy and low invasion. (author)

[en] A data compression method for MRI image sequences is developed by using the discrete cosine transform. The MRI images have the following two characteristics: they have few high-frequency components and their background image, i.e., outside of the regions of interest, has no significance. So considering these characteristics, we propose a new coding algorithm consisting of the following two methods: the bit-allocation method for a low-bit-rate compression based on the property of spectrum components and the coding method, with which the background is excluded from the coded object. We applied this algorithm to an actual MRI image sequence and obtained satisfactory results in compression ratio as well as image quality. (author)