Efficacy and Clinical Utility of a High-Attenuation Object Artifact Reduction Algorithm in Flat-Detector Image Reconstruction Compared With Standard Image Reconstruction

OBJECTIVE: To evaluate the effectiveness and clinical utility of a metal artifact reduction (MAR) image reconstruction algorithm for the reduction of high-attenuation object (HAO)-related image artifacts. METHODS: Images were quantitatively evaluated for image noise (noiseSD and noiserange) and qualitatively for artifact severity, gray-white-matter delineation, and diagnostic confidence with conventional reconstruction and after applying a MAR algorithm. RESULTS: Metal artifact reduction reduces noiseSD and noiserange (median [interquartile range]) at the level of HAO in 1-cm distance compared with conventional reconstruction (noiseSD: 60.0 [71.4] vs 12.8 [16.1] and noiserange: 262.0 [236.8] vs 72.0 [28.3]; P < 0.0001). Artifact severity (reader 1 [mean ± SD]: 1.1 ± 0.6 vs 2.4 ± 0.5, reader 2: 0.8 ± 0.6 vs 2.0 ± 0.4) at level of HAO and diagnostic confidence (reader 1: 1.6 ± 0.7 vs 2.6 ± 0.5, reader 2: 1.0 ± 0.6 vs 2.3 ± 0.7) significantly improved with MAR (P < 0.0001). Metal artifact reduction did not affect gray-white-matter delineation. CONCLUSIONS: Metal artifact reduction effectively reduces image artifacts caused by HAO and significantly improves diagnostic confidence without worsening gray-white-matter delineation.