.png)
Advancing medical imaging techniques to address the challenges in medicine
Methods for physical factor correction
Scatter constitutes a dominant physical factor that degrades image contrast and resolution in CBCT, and great efforts have been exerted by many researchers on developing efficient scatter correction methods. We have approached the scatter correction problem from various perspectives: scatter convolution kernel optimization, beam-blocker-based scatter estimation, and Monte-Carlo simulation approach. We have utilized the data consistency condition (DCC) in a circular CBCT to optimize the modeling parameters of the scatter convolution kernel, and demonstrated that the DCC effectively allows such an optimization (Phys. Med. Biol. 2015). We have developed a rebinned BPF algorithm that can successfully reconstruct image from partially blocked data while the blocked data can be used for scatter estimation and correction (Med. Phys. 2015;2019 - J. Min). The Monte-Carlo-based estimation of scatter has also been implemented in an accelerated parallel computing architecture (GPU) by modeling the tissue component from an initial reconstruction (Med. Phys. 2015 - K. Kim). We will be further investigating the developed technologies with an anti-scatter grid-based approach as well and will come up with the most efficient and reliable method for scatter correction.
Metal artifacts can overwhelm the image and need to be suppressed for preserving anatomical information in both medical and industrial applications. Depending on the primary cause of the MAR such as beam-hardening or photon starvation and on the degree of metal contamination, the performance of MAR methods may change drastically. Beam-hardening correction certainly helps the MAR as long as the artifacts are dominated by the polychromatic nature of the x-ray beams. Dual-energy scan can provide a solution to such a case as well. However, the metal artifacts mainly originating from photon starvation cannot be handled by either approaches. Sinogram in-painting method is a classical approach that can deal with the metal artifacts stemming from both beam-hardening and photon starvation. We have proposed a more accurate in-painting method than classical interpolation techniques that is based on iterative image reconstruction in the exterior-problem setting (J. Comp. Asst. Tomo. 2016). Its accelerated version that makes use of discrete algebraic reconstruction has also been developed and has been published (Nucl. Inst. Meth. A, 2016).
