摘要
We compared metal artifact in X-ray digital tomosynthesis (DT) and modern computed tomography (CT) reconstruction to improve the image quality. We compared the images of a prosthetic phantom (titanium) and a contrast-detail phantom obtained by DT using conventional filtered backprojection (FBP), metal artifact reduction (MAR) processing, and simultaneous iterative reconstruction technique (SIRT) methods and those obtained by CT using conventional FBP and adaptive statistical iterative reconstruction methods. The effectiveness of each method for enhancing the visibility of a prosthetic phantom was quantified in terms of the intensity profile and root mean square error, and the removal of ghosting artifacts was quantified in terms of the artifact spread function (ASF). In addition, low contrast resolution was evaluated in terms of the contrast-to- noise ratio. Image error was smaller in the MAR DT images in the near in-focus-plane, and the intensity profiles revealed the beam hardening effect. Streak artifacts were reduced in the SIRT DT and adaptive statistical iterative reconstruction CT images. The ASF performances of the algorithms were ranked in descending order: 1) MAR DT;2) CT (adaptive statistical iterative reconstruction, and conventional FBP);3) SIRT DT;and 4) conventional FBP DT. The low contrast resolution was higher in the CT images than in the DT images. In conclusion, a review of the results revealed that the metal artifact reduction was highest for tomosynthesis with MAR processing, and the low contrast resolution performance was highest for CT.
We compared metal artifact in X-ray digital tomosynthesis (DT) and modern computed tomography (CT) reconstruction to improve the image quality. We compared the images of a prosthetic phantom (titanium) and a contrast-detail phantom obtained by DT using conventional filtered backprojection (FBP), metal artifact reduction (MAR) processing, and simultaneous iterative reconstruction technique (SIRT) methods and those obtained by CT using conventional FBP and adaptive statistical iterative reconstruction methods. The effectiveness of each method for enhancing the visibility of a prosthetic phantom was quantified in terms of the intensity profile and root mean square error, and the removal of ghosting artifacts was quantified in terms of the artifact spread function (ASF). In addition, low contrast resolution was evaluated in terms of the contrast-to- noise ratio. Image error was smaller in the MAR DT images in the near in-focus-plane, and the intensity profiles revealed the beam hardening effect. Streak artifacts were reduced in the SIRT DT and adaptive statistical iterative reconstruction CT images. The ASF performances of the algorithms were ranked in descending order: 1) MAR DT;2) CT (adaptive statistical iterative reconstruction, and conventional FBP);3) SIRT DT;and 4) conventional FBP DT. The low contrast resolution was higher in the CT images than in the DT images. In conclusion, a review of the results revealed that the metal artifact reduction was highest for tomosynthesis with MAR processing, and the low contrast resolution performance was highest for CT.
