An Image Quality Assessment of a Metal Artifact Reduction Algorithm on a Novel C-Arm Cone-Beam CT System
Abstract
Purpose
To assess the image quality of cone-beam CT images that are reconstructed using metal artifact reduction algorithm on an C-arm linac.
Methods
A novel cone-beam imaging system on a C-arm linac was used to acquire images of spine phantom with titanium inserts. This phantom is custom made to mimic metal inserts in a patient. The images (here termed MAR iCBCT) are reconstructed using an iterative MAR algorithm. For comparison, the same phantom was scanned on a simulation, fan-beam CT scanner and the images (MAR CT) were reconstructed using the standard MAR algorithm. In addition, the phantom was also imaged on a previous version of the CBCT system, and the images (non-MAR CBCT) were reconstructed using a filtered back projection algorithm (FDK) which does not have MAR available. All image sets were rigidly registered to ensure spatial alignment for quantitative evaluation. Image quality was assessed using histogram analysis to compare the HU distributions.
Results
Visual inspection shows the MAR iCBCT does not show the metal artifacts present in the non-MAR CBCT, which exhibits streaking, blurring, and dark spots around the metal insert. On the other hand, the MAR iCBCT compares similarly to the fan-beam MAR CT image in terms of image quality. Histogram analysis shows that the intensity distribution of MAR iCBCT closely matches that of the MAR CT, particularly at the intensity peaks of high-density objects. However, the non-MAR CBCT shows a characteristically different distribution.
Conclusion
The iterative MAR algorithm for C-arm CBCT effectively reduces artifacts from titanium spinal inserts. Quantitative metrics suggest that this novel system provides image quality comparable to simulation fan-beam MAR CT. This shows promise for improving the accuracy of image-guided procedures in the presence of metallic hardware.