CT Reconstruction Using FBP with Sinusoidal Amendment for Metal Artefact Reduction

In this paper, a new method for metal artefact reduction in computed tomography is investigated, based on sinusoidal amendment. Each voxel in the scanned object corresponds to a sinusoidal curve in the projection data set. Some of the sinusoidal curves result from metal implant, which can be identified because the attenuation coefficient of the metal is much higher than that of human tissue, and therefore can be isolated from the projection, as well as reconstructed directly to provide the location of the metal implant. Then a highly accurate reconstruction image can be obtained using filtered back-projection because the original projection has been amended rather than interpolated. A real example is shown to demonstrate the method. 1 Introduction Since the attenuation coefficient of metal is much higher than human tissue, a metal implant causes low signal to noise, beam hardening and scatter, which cause the projection data to be corrupted, and bring star-burst artefact and error to the reconstruction image. The filtered back-projection (FBP) method is the most widely method for reconstruction of the scanning data since it gives high computational efficiency whilst keeping good accuracy [1], star-burst artefacts have been reduced by substituting the data corresponding to the projection lines through metal objects with the data from a neighborhood [2] or synthetic data using linear or polynomial interpolation strategies [3], or by adjusting the wavelet decomposition coefficients [4]. These methods are very effective for removing streaking generated by metal implants, and do not increase computational amount significantly. However, these strategies inevitably cause some information loss and can still result in artefacts. Iterative algorithms have been regarded as a potential method of providing high quality CT reconstruction images, especially for metal artefact reduction [5-8]. Using an iterative algorithm, the data affected by the metal implant, instead of being replaced, is ignored. Computational expense was the main problem preventing the iterative method from practical clinical application, and much effort has therefore been focused on this subject [9][10]. An alternative strategy to avoiding metal artefact is by using low attenuation material such as Titanium [11][12]. However, this is not ideal for clinical