Quantitative phantom-based comparison of commercial CT metal artifact reduction algorithms

Area of Science:

• Medical Imaging
• Radiological Physics
• Computational Imaging

Background:

• Metal artifacts in computed tomography (CT) degrade image quality and CT number (CTN) accuracy.
• This compromise impacts downstream applications like radiation therapy planning.
• Limited standardized methods exist for quantitatively comparing commercial metal artifact reduction (MAR) algorithms.

Purpose of the Study:

• To quantitatively compare the performance of four commercial MAR algorithms.
• To utilize standardized metrics and a novel analytical approach for evaluation.
• To provide vendor-neutral performance data for MAR algorithms.

Main Methods:

• A custom resin phantom with various metal inserts (aluminum, titanium, stainless steel) was scanned on four CT systems.
• Proprietary MAR algorithms from Canon, GE, Philips, and Siemens were applied.
• Quantitative analysis included circular profile analysis, geometric accuracy assessment, and volumetric artifact characterization using a composite error metric (M-error index).

Main Results:

• All MAR methods showed increased CTN deviation near metal interfaces.
• Single-energy metal artifact reduction (SEMAR) and iterative metal artifact reduction (iMAR) demonstrated the most consistent geometric accuracy (<5%) across all metals.
• O-MAR exhibited material-dependent performance with higher errors for aluminum and stainless steel, while Smart-MAR showed significant geometric errors (>19%) for titanium and stainless steel.

Conclusions:

• SEMAR and iMAR provide the most reliable geometric accuracy and artifact suppression for diverse metal implants.
• O-MAR's performance is inconsistent across different metal materials.
• Phantom-based benchmarks offer valuable, controlled data for evaluating MAR algorithm efficacy.