A Cone Beam CT Scatter Reduction Method with Piecewise Prior Image Knowledge
X Li*, T Li, Y Yang, D Heron, M Huq, University of Pittsburgh Medical Center, Pittsburgh, PATU-A-213CD-6 Tuesday 8:00:00 AM - 9:55:00 AM Room: 213CD
Purpose: In cone beam computerized tomography (CBCT), the reconstructed image contains significant image artifacts induced by scatter effects. The purpose of this study is to develop a scatter reduction method to improve CBCT image quality with piecewise prior image knowledge.
Methods: The hypothesis of the proposed method is that in scatter-free CBCT image, the image intensities of different kind of materials are approximately globally uniform, i.e. piecewise property. However, the scatter induced image artifacts diminish the piecewise property of the CBCT image. In the proposed scatter reduction method, an extended fuzzy C-mean (EFCM) function is used to describe the degree of piecewise property of the CBCT image. Since the scatter components in projection domain can be characterized as low frequency, by minimizing the EFCM function in image domain, the proposed method removes an appropriate low frequency component from projection domain so that the reconstructed CBCT image achieves a piecewise property.
Results: The performance of the proposed scatter reduction algorithm was tested using both digital phantom and pelvis CBCT images. In a phantom study, the proposed method reduced the maximum HU errors from 420HU to about 80HU; in a pelvis case, the HU errors were reduced from 580HU to 150HU. All those maximum HU errors occur around area with high image contrast. For soft tissues, the proposed method reduced the HU errors to less than 60HU. In addition, there are severe shading-artifacts that appeared in the uncorrected CBCT images in which the proposed method significantly improved the image intensity uniformity of different tissue types in the corrected CBCT image.
Conclusions: For an object with piecewise property, the proposed scatter reduction method demonstrated substantial reductions in scatter-induced image artifacts commonly encountered in CBCT imaging. This methodology may be an avenue to significantly improve the image quality of clinical CBCT images.