Cancer is the leading cause of death in the world. In order to improve the survival rates against the cancers, early detection is essential. Recently, positron emission tomography/computed tomography （PET/CT） is being used for cancer screening in Japan and some Asian countries. However, it generates about 1000 slice images per scan; the computer‐aided diagnosis （CAD） is expected to keep the diagnostic accuracy. Our research group develops the CAD scheme for multimodality and multi‐organ using PET/CT images. This review article describes our progress and future works in the developments of CAD scheme using PET/CT images.

抄録全体を表示

Digital imaging have many advantages such as wide dynamic range, post processing, multiple viewing options. The CDRS which creates the reference image using the post processing technology of two digital radiographs, is useful to inexperienced radiologists. This CDSR makes it easier to compare multiple radiological images having different anatomical or functional information with the same spatial co‐ordinates. On the other hand, in digital X‐ray imaging systems, good images are obtained for a large range of doses, so a tendency to use higher patient doses than necessary could occur. IEC specifies definitions and requirements for the exposure index（EI）of images acquired with digital X‐ray imaging systems. The advantages and problems of such digital X‐ray imaging are considered.

抄録全体を表示

Dynamic chest radiography with computer analysis is expected to be a new type of functional imaging system, which can quantify and visualize cardiopulmonary function on dynamic chest radiographs, such as diaphragm motion, heart wall motion, pulmonary ventilation, and blood circulation. We will first introduce the backgraound, imaging methods, and then image analysis methods for each evaluation item. We pay particular attention to interframe subtraction and mapping technique, which play a critical role in the evaluation of pulmonary ventilation and blood circulation. We also discuss features, future perspectives, and issues related to dynamic chest radiography on the basis of preliminary clinical study.

抄録全体を表示

Image recognition/processing results would be useful to physicians, surgeons, nurses and radiological technologists for improvement of the quality of medical care and prevention of malpractice in their daily work. In fact, previous studies indicated the usefulness of the computerized method for discovering misfiled chest radiographs with a high performance. With image recognition/processing results, radiologists can minimize imaging adjustments and radiologists decreased reading time without compromising accuracy or confidence. Therefore, we have introduced several methods for improvement of the quality of medical care and prevention of malpractice: the automated Classification method of the Body Parts in Digital Radiographs, the position check support system to identifies and highlights lines and tubes on chest X‐ray images, the preview system for prevention of X‐ray miss examination, automatic MPR processing from head volume CT data.

抄録全体を表示

Cerebrovascular diseases are the fourth leading cause of death in Japan. Therefore, a screening system called the Brain Check‐up is widely employed. Although the number of cerebrovascular diseases has gradually decreased every year, the number of case of cerebral infarction and subarachnoid hemorrhage are on the increase. Therefore, it is important to reduce the incidence of these diseases. For assisting radiologists in the early detection of cerebrovascular diseases, we developed several computer‐aided diagnosis（CAD）schemes. This review article describes two CAD schemes, which are for the detection of lacunar infarct and the automatic segmentation of leukoaraiosis. The detection of lacunar infarcts is important because their presence indicates an increased risk of severe cerebral infarction. However, accurate identification is often hindered by the difficulty in distinguishing between lacunar infarcts and enlarged Virchow‐Robin spaces. Thus, CAD scheme for the detection of lacunar infarcts would be helpful for radiologists' diagnosis. The quantitative analysis of leukoaraiosis is also important because its presence and extension is associated with an increased risk of severe stroke. However, the diagnosis of leukoaraiosis has generally been limited to subjective judgments by radiologists. Therefore, a computerized method providing an objective measurement of the lesion volume would be useful for radiologists' image interpretation. In this review article, we describe the current status of development of these CAD schemes and the future direction of the research.

抄録全体を表示

We have been developing a computer‐aided diagnosis （CAD） system for hypertensive retinopathy, diabetic retinopathy and glaucoma. This paper describes recent our retinal CAD studies. To detect the arteriolar narrowing, which is the finding of hypertensive retinopathy, the performance of the arteries and veins classification was improved by extracting the only veins on the red bit component of the color retinal images. We also improved the detection of microaneurysms, which is the finding of diabetic retinopathy, by combining three detectors, which were the double‐ring filter, Shape Index based on Hessian matrix, and Gabor filter. Moreover, we improved the measurement of rim‐to‐disc ratio, which is glaucoma finding, by addition of the blood vessel bends detection. In the next stage, we plan to develop the difference analysis between the previous retinal image and current one for observation of the treatment effects of the metabolic syndrome. We also plan to develop the CAD system for the retinal images with wide‐field scanning laser.

抄録全体を表示

This article reviews our recent studies on the fully automatic recognition of the skeletal muscles in non‐contrast torso CT images. The aging rate of our country is the world's highest level. Therefore, the extension of healthy life expectancy is one of the most important issues. In particular, the maintenance and increase of muscle volume have been included as a national policy in “Healthy Japan 21". Considering the extension of healthy life expectancy, it is not limited to old age, skeletal muscle mass as health promotion from younger age groups is important. In this paper, first, the background of the skeletal muscle recognition in non‐contrast CT images is described. Then the research results obtained by two projects, “Future CAD" and “Computational Anatomy" projects are shown. Finally, the future prospective of the research on the skeletal muscle recognition is described.

抄録全体を表示

The Arterial disease caused by lifestyle, and mortality is 30% or more. Atherosclerosis is contributed not only medical factor, but also the stress by hemodynamics. The aim of this study was to evaluate the impact of stress in the carotid artery using computed fluid dynamics analysis. We made computational fluid dynamics analysis system by open sourse softwear. And we evaluated stress and flow velocity using a carotid artery model that was based MRA images. The carotid bifurcation analysis model showed a voltex flow and a decrease in flow velocity in the carotid bulb area. The finding results suggested that hemodynamics contributes to the atherosclerosis.

抄録全体を表示

In recent years in diagnostic radiology, the number of multimodality images as well as available supplemental information, including images with known diagnosis in the database, is becoming larger and larger. Therefore it is imminent to have a system that can efficiently and correctly gather the useful information as diagnostic reference. We have been investigating similarity measures and methods for similar image retrieval of breast lesions on mammograms. In this paper, a brief background and our current investigation on the method are introduced. In the second half, our recent investigation on image analysis of dental panoramic radiographs（DPRs）is introduced. Computer-aided diagnosis on DPRs can promote a supplemental screening of systemic diseases through dental clinics where DPRs are routinely obtained for dental examination purposes.

抄録全体を表示

The optically stimulated luminescence（OSL）dosimeter is a one of the useful detectors to measure exposed dose, and the information of dose is absolutely imperative to analyze the medical images. NanoDot dosimeter is a small‐type OSL dosimeter, 10mm×10mm×2mm, commercialized by Landauer, Inc. This dosimeter can be read by a microStar, in which weak LED light is exposed to the detector and then emitted luminescence is detected. Because the LED light is enough weak, the dosimeter can be read repeatedly. In this paper, we proposed the high accuracy measurements by consecutive readings. In order to verify our procedure, we used the diagnostic X‐ray equipment. Our experiments were carried out in the following conditions in which a minimum absorbed dose is assumed to be about 0.05 mGy or less: tube voltage of 80kV, tube current‐time product of 0.5mAs to 500mAs, and X‐ray source to detector distance of 1m. The dosimeters are read consecutive 100times, and then based on these data we discuss about the relationship between the measurement accuracy and repeated times. Then we concluded that the dosimeter should be read about 10times or less to reduce the error to a few percentages. Our procedure using nanoDot is available to estimate the absorbed dose in the clinical site.

抄録全体を表示

To perform investigation of image‐quality improvement in photon‐counting energy‐dispersive X‐ray computed tomography（ED‐CT）under low count rates, we have developed an ED‐CT system using a frequency‐voltage converter（FVC）and a cadmium telluride detector. The FVC is used to improve the image granulation by compensating for the photon‐count fluctuation. In the ED‐CT, the photon energy range is regulated using a multichannel analyzer（MCA）by determining both the maximum and minimum energies. The FVC consists of a microcomputer, an integrator, and an operational amplifier. When 0.2‐μs‐width event logical pulses from the MCA are sent to the microcomputer, 5‐μs‐width 5.0‐V‐height logical pulses are produced from the microcomputer and sent to the integrator to produce the time‐average voltage which is proportional to the count rate. The integrator output is then amplified by the amplifier, and the FVC output is input to an analog‐digital converter. Pileup‐less CT was accomplished by repeated linear scans and rotation of the object at a tube voltage of 70 kV and a count rate of 1.9 kilo‐counts per second or below. We also performed the iodine K‐edge CT and the energy subtraction.

抄録全体を表示

The detection of lacunar infarcts is important because their presence indicates an increased risk of severe cerebral infarction. However, their accurate identification is often hard because of the difficulty in distinguishing between lacunar infarcts and enlarged Virchow-Robin spaces. Therefore, we developed computer-aided diagnosis scheme for the detection of lacunar infarcts. The performance of our previous method indicated that the sensitivity of 96.8% with 0.76 false positive（FP）per slice. However, further reduction of FPs was remained as an issue to be solved for the clinical application. In this paper, we proposed AdaBoost template matching. This classifier can distinguish between lacunar infarcts and FPs by selecting suitable templates in the template matching. By using this technique, 55.5% FPs were eliminated while keeping the same sensitivity. Thus the proposed method was found to be useful for the sophistication of the automatic detection of lacunar infarcts in MR images.

抄録全体を表示