At the National Institute of Radiological Sciences, we are developing OpenPET, an open-type positron emission tomography (PET) geometry with a physically open space, which allows easy access to the patient during PET studies. Our first-generation OpenPET system, dual-ring OpenPET, which consisted of two detector rings, could provide an extended axial field of view (FOV) including the open space. However, for applications such as in-beam PET to monitor the dose distribution in situ during particle therapy, higher sensitivity concentrated on the irradiation field is required rather than a wide FOV. In this report, we propose a second-generation OpenPET geometry, single-ring OpenPET, which can efficiently improve sensitivity while providing the required open space. When the proposed geometry was realized with block detectors, position-dependent degradation of the spatial resolution was expected because it was necessary to arrange the detector blocks in ellipsoidal rings stacked and shifted relative to one another. However, we found by Monte Carlo simulation that the use of depth-of-interaction (DOI) detectors made it feasible to achieve uniform spatial resolution in the FOV.

View full abstract

High-dose radiation destroys cancer cells in radiation therapy. Portal imaging for geometric verification in image guided radiation therapy is performed to adjust the patient's position to deposit the prescribed dose correctly within the prescribed tumor region. However, conventional portal images have lower image contrast than diagnostic X-ray images because portal imaging uses a therapeutic high-energy X-ray beam. We have applied a refraction contrast technique to portal imaging to improve the accuracy of radiation therapy. The edge enhancement effect in refraction contrast portal imaging can provide images with higher contrast than those obtained by conventional portal imaging. In the present study, head images of rats were obtained using refraction contrast portal imaging and conventional portal imaging. The images were then compared to assess the degree of improvement in image quality.

View full abstract

In order to elucidate the process of morphological formation in the development of human embryos, it is useful to study three-dimensional (3D) images. We performed 3D reconstruction of serially sectioned images from a massive collection of human embryo specimens at Kyoto University that had been acquired with no preparation for 3D reconstruction. For such massive collections of sectioned specimens, it is difficult to avoid problems such as deficits, damage, and processing errors. Therefore, the reconstruction procedure was performed with these problems taken into consideration. We have also developed a system for retrieving and presenting such 3D reconstructed images.

View full abstract

This paper presents a new method for automated SmartPrep tracker positioning in liver MRI scans. SmartPrep is used to monitor the contrast bolus signal in order to detect the arrival time of the bolus. Accurately placing the tracker in the aorta while viewing three planar scout images is a difficult task for the operator and is an important problem from the workflow standpoint. The development of an automated SmartPrep tracker would therefore help to improve workflow in liver MRI scans. In our proposed method, the aorta is detected using AdaBoost (which is a machine learning technique) by searching around the CSF in the spinal cord. Analysis of scout scan images showed that our detection method functioned properly for a variety of axial MR images without intensity correction. A total of 234 images reconstructed from the datasets of 64 volunteers were analyzed, and the results showed that the detection error for the aorta was approximately 3 mm.

View full abstract

To solve the interior problem in computed tomography, a priori knowledge of the sub-region located inside the region of interest (ROI) is usually required. However, this is somewhat paradoxical because the a priori knowledge of the sub-region is located inside the ROI that we want to obtain through reconstruction. Here we show that the interior problem has a unique solution even if the a priori knowledge of the sub-region is located outside the ROI when lines called exact lines exist and satisfy the following two conditions. The first condition is that the exact line region should belong to the a priori knowledge region or ROI, and the second condition is that overlap between the exact line region and ROI always exists. By using a property of the Hilbert transform, the part of the object function ƒ(x) located inside the ROI can be obtained from a priori knowl-edge located outside the ROI. As a result, we can obtain new a priori knowledge located inside the ROI, and thus existing interior problem methods can be used to generate the reconstructed image. This paper is an extension of the work reported in "Tiny a priori knowledge solves the interior problem in computed tomography", which was published in Physics in Medicine and Biology in 2008. The new results presented here indicate that under relaxed constraints, the a priori knowledge of a sub-region outside the target ROI can provide exact and stable reconstruction from interior truncated projections. The results of experiments performed using simulated data have shown the validity of the proposed method.

View full abstract

Conventional machine learning-based segmentation (e.g., ensemble learning) suffers from the problem of unnatural shapes of the extracted figures because decision-making by the constructed classifier is carried out voxel by voxel or local region by local region independently. In this paper, we propose an ensemble learning algorithm that constructs a segmentation process based on the statistical shape feature of an organ. We describe a novel loss function for evaluating the shape of an extracted figure using a statistical shape model of the organ and an algorithm to minimize the loss function which combines conventional error loss with proposed loss. The results of experiments using an artificial image are presented to confirm the basic performance of the algorithm. In addition, the results of experiments involving spleen segmentation using 80 clinical CT volumes are presented to validate the clinical usefulness of the algorithm. Based on these results, it is concluded that the proposed algorithm reduces unnatural shapes of the extracted organs and provides significantly superior segmentation performance as compared to conventional ensemble learning-based segmentation.

View full abstract

It is difficult to evaluate human soft tissues in detail using conventional X-ray absorption imaging due to the small density differences. By analyzing the phase shifts based on the inherent characteristics of X-ray waves, it is possible to obtain images of soft tissues with approximately 1000 times higher sensitivity as compared to the conventional method employing X-ray absorption. Using a phase-contrast X-ray CT system with an X-ray interferometer, we obtained images of rat hearts fixed in either formalin or ethanol. After each rat was anesthetized, it was perfused with physiological saline solution and fixed with either formalin or ethanol. The heart was surgically resected and replaced in each solution, and phase-contrast X-ray CT imaging was then performed at KEK. Formalin fixation has traditionally been used to prepare tissue samples, but we employed ethanol fixation to further enhance image contrast. In phase-contrast X-ray CT images, three myocardial layers (the endocardium, midcardium, and epicardium) were clearly depicted in the ethanol-fixed hearts, but these myocardial structures were not visible in the formalin-fixed hearts. Observation of this myocardial layer structure makes it possible to analyze abnormalities such as injury to specific myocardial layers, which previously could not be visualized.

View full abstract

The uniform flip angle throughout the imaging region is the requirement to the RF transmit subsystem. It started as solenoid and saddle coils, followed by the slotted tube resonator. And the birdcage coil, which produces very uniform RF field with the quadrature excitation, was commonly accepted as the solution. However, the wavelength effect and high SAR due to the increased strength of the static magnetic field are requiring further evolution of the RF transmit system now. This paper reviews the basic technique of MRI transmission and solution in high field MRI.

View full abstract