[en] Chest radiography is the most widely used imaging modality in medicine. However, the diagnostic performance of chest radiography is deteriorated by the anatomical background of the patient. So, dual energy imaging (DEI) has recently been emerged and demonstrated an improved. However, the typical DEI requires more than two projections, hence causing additional patient dose. The motion artifact is another concern in the DEI. In this study, we investigate DEI-like bone-suppressed imaging based on the post processing of a single radiograph. To obtain bone-only images, we use the artificial neural network (ANN) method with the error backpropagation-based machine learning approach. The computational load of learning process of the ANN is too heavy for a practical implementation because we use the gradient descent method for the error backpropagation. We will use a more advanced error propagation method for the learning process

[en] X-ray defect inspection apparatus can be used in the production line to inspect the PCB. However, a simple X-ray radiography cannot discriminate defects from the multi-layer PCBs because the layers of them overlays the defects. To complement this issue, computed tomography (CT) technology is applied to the NDT system which can offer 3-dimensional information of object. However, CT requires hundreds of projection images to examine a single PCB, hence real-time inspection is nearly impossible. In this study, we develop a planar computed tomography (pCT) system appropriate for the multi-layer PCB inspection. For the image reconstruction of planar cross-section images, we use the digital tomosynthesis (DTS) concept in association with the limited angle scanning. and performance characterization of the pCT system for the PCB inspection. The 3-d Fourier characteristics and more quantitative performance, such as contrast, uniformity, depth resolution will be presented. The cross-sectional images of multi-layer PCBs will also be demonstrated

[en] It is possible to use the method of scanning the object with continuous motion. This method has the advantage of taking less time to scan, but it has a disadvantage in the quality of acquired image that is deteriorated. In order to select a suitable scanning method for the requested task, the quantitative characteristics evaluation of the two methods of scanning is required in the μ-CT system. The purpose of this study is to investigate the signal and noise characteristics of continuous scanning μ-CT system experimentally. In addition, we compare these with the characteristics of step motion scanning. Due to continuous rotation movement of mouse, the tomographic image acquired by continuous motion scanning was blurrier than that by step motion scanning.

[en] The purpose of this study is to compare the quality of images obtained from the step-and-shoot and continuous scanning methods in terms of system resolution. In addition, this study aims to determine the fastest continuous scanning speed with image quality enough for the defect inspection. Computed tomography (CT) is an emerging technology for the non-destructive testing (NDT) of high precision manufacturing to enhance the production reliability. In order to use the CT for the NDT, high-speed inspection is essential for a higher production yield. However, conventional high resolution CT systems usually require terms of minutes to take exam because they employ the step-and-shoot protocol to avoid motion blur during the image acquisition. Unlike the step-and-shoot protocol, continuous scan protocol, which simultaneously acquires projection images during the rotation, can only require terms of seconds because it can ignore the time delay between 'stop and go' steps of rotation. Continuous scan is used for medical CT systems. However, the loss of system resolution originated from the motion blur cannot be avoidable. In order to investigate the relationship between acquisition time and image quality of continuous scanning, the experiment will be conducted with varying rotational speed of object