Introduction: Metal artefacts are a common clinical challenge when assessing CT images of patients with metal implants. This is because metallic implants markedly degrade image quality due to excessive attenuation of the X-ray beam. Factors influencing the level of metal artefacts are the metal composition of the metal implant, the acquisition parameters, the image reconstruction and the image reconstruction algorithm. Purpose: The purpose of this thesis is to compare the performance of different algorithms for the removal of artefacts in CT images caused by the presence of metal in the scan region. We will compare how the presence of artefacts is reduced in virtual monoenergy reconstructions and how it is reduced in reconstructions where the MAR algorithm is used. Methods: From the PACS database of the Faculty of Health Sciences in Ljubljana, we selected four cases showing the spine, hip, knee and ankle with various metal plates and screws implanted which represent artefacts on the CT images. We then created VME reconstructions, with different kernels and combinations of different filters, in the range of 40-190 keV and three classical reconstructions that are used on a daily basis in the clinical setting. Five evaluators first looked at all the VME reconstructions and chose their favourites among all of them. Then, the classical reconstructions were added to these selected series and the evaluators chose the final favourite or the most suitable for clinical use. Results: We found that the combination of VME reconstructions and the MAR algorithm with Br59 kernel for bone and Br36 for soft tissue and a range of energies between 140 and 160 keV, is the most optimal for the assessment of both bone and soft tissue structures. Discussion and conclusion: Together with the studies already done, we agree that the combination of VME and MAR reconstructions in combinations with higher energies is the most optimal for metal artefact reduction and diagnostic use.
|
