Overviews for Donghyeon's previous research topics
High accuracy photon-counting detector model
Published in Medical Physics (2023).
Deep learning based energy-flexible reconstruction algorithm
Published in arXiv (2023).
One-step simultaneous algebraic reconstruction technique (GSART) for dual-energy CT
Clustering algorithm for noise reduction in dual-energy X-ray cargo inspection system
We developed an effective clustering method designed to mitigate noise, which is considerably amplified during material decomposition processes. Many clustering algorithms in imaging process tend to create artificial patterns on the images they are applied to. However, our proposed method incorporates the proposed probabilistic regularization to overcome the issue, enabling the combination of similar clusters and the effective elimination of very small clusters. This approach not only enhances the overall accuray of material decomposition but also maintains resolution with small compromise.
Published in Nuclear Inst. and Methods in Physics Research, A (2020).
Material decomposition method using a rotating cylindrical phantom in dual-energy X-ray cargo inspection system
This research introduces an innovative approach for material decomposition in dual-energy X-ray cargo inspection system. Unlike the conventional calibration phantoms that are typically large and heavy to encompass possible thickness ranges of inspected materials, the proposed calibration phantom allows to easily measure various and extensive thickness combinations of low- and high-density materials through a simple rotation around its center. Our findings indicate that this approach allows for highly accurate material decompsition across various materials and thicknesses. The figure shows the proposed calibration phantom and the material decomposed images.
Published in Nuclear Inst. and Methods in Physics Research, A (2018).
Single-scan dual-energy cone-beam CT using a multi-slit moving filter
This research aimed to introduce a novel method for dual-energy CT imaging. It involves the use of a multi-slit moving filter positioned in front of the X-ray source. This setup allows to generate a high-energy X-ray spectrum by hardening the X-ray source spectrum, while also achieving the necessary sampling density crucial for image reconstruction of imaged objects. The figure illustrates the experimental system for the proposed approach and an example sinogram. Our findings confirm that high-quality dual-energy CT images can be produced using this method, with the added benefit of significantly reduing the radiation dose.