- 期刊
Efficient Metal Artifact Reduction Method Based on Improved Total Variation Regularization
並列摘要
Metal implants produce strong artifacts in reconstructed computed tomography images and thus severely reduce image quality. This study proposes a method for metal artifact reduction that combines modified compressed sensing reconstruction and the sinogram inpainting method. The procedure starts with total variation reconstruction to obtain an initial image. Then, the image is recovered using improved total-variation-based regularization. This process produces an artifact-free background image. The missing information in the original sinogram is complemented using the forward projection of the background image. Consequently, the transitions between the original sinogram and artificial sinogram are smoothed by re-matching the baseline. The algorithm is validated using simulations and phantom data. Results show that streaks are eliminated and shadows are significantly reduced.
延伸閱讀
- Romero, J. R., Carballido, J. A., Garbus, I., Echenique, V. C., & Ponzoni, I. (2016). A Bioinformatics Approach for Detecting Repetitive Nested Motifs using Pattern Matching. Evolutionary Bioinformatics, (2016), 247-251. https://doi.org/10.4137/EBO.S40138
- Gao, Y. L., & Liu, J. M. (2012). Multiobjective Differential Evolution Algorithm with Multiple Trial Vectors. Abstract and Applied Analysis, 2012(), 916-927. https://doi.org/10.1155/2012/172041
- Lin, C. A. (2012). An Efficient Temporal Model for Action Recognition Using Multivariate Linear Prediction [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU.2012.00105
- Marquez, A., Oviedo, J. J. E., & Odloak, D. (2013). Model Reduction Using Proper Orthogonal Decomposition and Predictive Control of Distributed Reactor System. Journal of Control Science and Engineering, 2013(), 254-272. https://doi.org/10.1155/2013/763165
- Wang, Y., & Zhu, L. (2016). Variable Selection and Parameter Estimation with the Atan Regularization Method. Journal of Probability and Statistics, 2016(), 127-138. https://doi.org/10.1155/2016/6495417