Metal Artifact Reduction in Chest CT of Patients with Cardiac Implantable Devices

Objectives: To determine whether smart metal artifact reduction (SMAR) combined with adaptive statistical iterative reconstruction (ASIR) improves thoracic computed tomography (CT) image quality in patients with cardiac implantable electronic devices (CIEDs) compared with ASIR alone. Materials and Methods: In this retrospective single-center study (January 2022-December 2024), 30 consecutive CIED carriers (mean age 63.4±14.5 years; 18 pacemakers, 12 implantable cardioverter-defibrillators) who underwent thoracic CT on a 512-slice scanner were evaluated. Paired ASIR-only and ASIR+SMAR reconstructions were generated from identical raw data. Two blinded radiologists graded image quality on a 5-point scale across soft-tissue, osseous, and pulmonary windows at proximal (<3 cm), intermediate (3-5 cm), and distal (>5 cm) zones. A third radiologist measured quantitative artifact load as the Hounsfield unit standard deviation (HU-SD) in proximal-zone regions of interest. Results: ASIR+SMAR significantly reduced proximal-zone HU-SD compared with ASIR alone (43.60±17.33 vs. 90.40±42.11; p<0.001; Cohen’s d=1.42), representing a 52% reduction. Overall qualitative scores improved significantly (3.84±0.48 vs. 2.92±0.47; p<0.001). The greatest gains occurred on the soft-tissue windows in the proximal and intermediate zones (p<0.001). However, algorithm-related streaks appeared in distal lung parenchyma in 13 of 30 patients (43.3%), with 5 (16.7%) being severe enough to preclude evaluation. Inter-reader agreement was substantial (weighted κ: 0.68-0.72). Conclusion: SMAR combined with ASIR significantly improves image quality metrics in the proximal and intermediate zones around CIED generators. However, algorithm-induced artifacts in distal lung parenchyma occur in approximately 43% of patients. A dual-reconstruction review strategy is recommended for clinical practice.