2233 - User Evaluation of Metal Artifact Reduction Reconstruction for Daily Radiotherapy Treatment Imaging

A. Cetnar, M. Ramani, R. S. Handschuh, D. J. DiCostanzo, and N. Gupta; The Ohio State University, Columbus, OH

Purpose/Objective(s): The goal of this study is to evaluate clinician preference of cone beam computed tomography (CBCT) imaging with Metal Artifact Reduction (MAR) vs. standard (FDK) reconstruction algorithms with common implants for a standard c-arm linear accelerator with novel CBCT technology. Materials/

Methods: A survey was developed with sets of MAR and FDK images displayed side-by-side with the same window/level in randomized order. Images included four different fiducial markers embedded in an anthropomorphic phantom, a spinal fixation device and a hip implant placed in a water phantom, and an electron density phantom with a titanium insert. The survey was completed by physicians, physicists, dosimetrists, and therapists at an academic cancer hospital. Respondents selected which of the two images had superior image quality based on better ability to distinguish the metal implant and less surrounding artifact. Test reliability was conducted by repeating a subset of randomized images.
Results: In the completed surveys, the MAR algorithm was preferred 99% of the time for the electron density phantom, 99% for a spine fixation device, and 86% for the hip implant. The preference for fiducial markers had more variation with MAR preferred 57% for a carbon fiducial, 55% for a cylindrical gold marker, 37% for a gold chain, and 27% for an S-shaped platinum fiducial. Summary of results from clinician roles is shown in Table 1. Using test-retest theory for the questions repeated during the survey, the Pearson Correlation Coefficient was 0.989 indicating very strong reliability. The MAR algorithm reduced the visible streaking artifacts for spinal fixation device, hip implants, titanium rods, and for gold fiducial markers compared to the standard reconstruction algorithm. Greater streaking artifact was observed in the MAR than standard reconstruction for the other fiducial markers.
Conclusion: By comparing images of commonly used metal implants, we observe that the MAR algorithm is the preferred image for clinicians for hip implants and spinal fixation devices. Greater variability was shown in the preference between the algorithms among fiducial markers. Abstract 2233 – Table 1

	Electron Density Phantom	Hip Implant	Spine Hardware	Gold Cylindrical Fiducial	Gold Chain Fiducial	Platinum S-shaped Fiducial	Carbon Fiducial
All Clinicians (N=43)	99%	86%	99%	55%	37%	27%	57%
Physicians (n=13)	100%	90%	100%	62%	46%	23%	62%
Physicists (n=13)	100%	92%	100%	46%	23%	18%	46%
Therapists (n=12)	96%	78%	96%	58%	42%	39%	53%
Dosimetrists (n=5)	100%	80%	100%	53%	40%	33%	80%