Dose calculation accuracy with extended CT scales near metal: phantom-patient evaluation using a beam-metal overlap metric.

BACKGROUND: We evaluated whether 16-bit CT reconstruction with explicit CT-to-electron-density (CT-to-ED) calibration improves dose-calculation accuracy versus conventional 12-bit reconstruction in the presence of metallic implants, and whether geometric metrics can identify cases that benefit most. METHODS: CT-to-ED tables were built for both bit depths. In Solid Water, dose profiles and absolute dose were compared against EBT4 film and TG-51 ion-chamber measurements. Eight clinical IMRT plans were optimized and cross-recalculated between bit-depth domains with segments and monitor units fixed. A beam–metal interaction metric, ηmetal (monitor-unit–weighted beam’s-eye-view overlap), and the volumetric overlap ratio (VOR) between metal and PTV were computed. Gamma passing rate (GPR, 3 mm/3%), mean gamma, profile error, and absolute-dose deviation were evaluated. RESULTS: In the phantom, 16-bit improved agreement over 12-bit, yielding higher GPR, profile differences ≤ 2%, and absolute-dose errors ≤ 1.0%; 12-bit showed profile errors up to 5.4% (stainless steel) and 3.0% (titanium) and absolute-dose errors of 2.3% and 1.7%. In patients, ηmetal > 35% was associated with GPR < 90% when 12-bit plans were recalculated on 16-bit, whereas low ηmetal maintained ≥ 90%. ηmetal correlated positively with VOR and negatively with GPR, supporting pre-treatment triage via VOR. CONCLUSIONS: Extended-bit reconstruction mitigates HU saturation, improves RED assignment, and yields better agreement with measurements and calculation. These findings support adopting 16-bit CT with CT-to-ED calibration as the default for planning and verification in patients with metallic implants, with VOR or ηmetal flagging cases where benefit is greatest.