Tomosynthesis is a useful imaging tool for breast, lung, and orthopedic diagnostics. Compared with computed tomography (CT), fewer artifacts are caused by metal components (metal artifacts). This advantage makes tomosynthesis particularly useful for orthopedics. Implementing filtered back projection (FBP) with a modified kernel leads to an increase in the low-frequency components of reconstructed images and reduces metal artifacts in tomosynthesis. However, even using this reconstruction method, metal artifacts are present in the region very close to any piece of metal. Due to the modified kernel, the observation of fine structures is difficult. We developed a new reconstruction algorithm to provide fewer metal artifacts in tomosynthesis images than in conventional images without filtering. Our new algorithm consists of four steps: 1) automatically extracting metal components from projection images using a novel method that we developed; 2) dividing projection images into metal-free projection images and metal-only projection images; 3) reconstructing these two projection images using the ordered subset-expectation maximization (OS-EM) method to create metal-free tomosynthesis images and metal-only tomosynthesis images; and 4) combining the tomosynthesis images and thereby obtaining metal artifact-reduced tomosynthesis images. Our new metal extraction method in step 1 is based on the graph cuts algorithm. We compared four image reconstruction algorithms: (a) FBP, (b) FBP with a modified kernel, (c) simple OS-EM and (d) the proposed method. The results demonstrate that the proposed method significantly reduces metal artifacts when compared with the other methods.
[1]
Tokuo Umeda,et al.
Evaluation of the X-ray digital linear tomosynthesis reconstruction processing method for metal artifact reduction
,
2009,
Comput. Medical Imaging Graph..
[2]
Marie-Pierre Jolly,et al.
Interactive graph cuts for optimal boundary & region segmentation of objects in N-D images
,
2001,
Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.
[3]
H. Malcolm Hudson,et al.
Accelerated image reconstruction using ordered subsets of projection data
,
1994,
IEEE Trans. Medical Imaging.
[4]
Fucang Jia,et al.
Cone-Beam Computed Tomography Image Reconstruction Based on GPU
,
2012
.