Harmonization of technical image quality in computed tomography: comparison between different reconstruction...

Area of Science:

• Medical Imaging Physics
• Radiological Technology

Background:

• Computed tomography (CT) scanners utilize diverse, proprietary reconstruction algorithms and kernels.
• Ensuring consistent image quality across different CT scanners is a significant challenge in radiology.
• Harmonizing technical image quality is crucial for reliable diagnostic interpretation and multi-center studies.

Purpose of the Study:

• To modify and evaluate a reconstruction algorithm and kernel matching scheme for CT image quality harmonization.
• To address the increasing difficulty of maintaining consistent image quality between CT scanners with varying reconstruction methods.
• To develop a strategy for harmonizing technical image quality across different CT scanners.

Main Methods:

• Utilized the Catphan 600 phantom scanned on six CT scanners from four vendors.
• Acquired data at different CT dose indices (10 mGy and 40 mGy) and slice thicknesses (1 mm and 5 mm).
• Reconstructed data using various algorithms and kernels, applying a modified matching scheme based on noise power spectrum (NPS) and modulation transfer function (MTF).

Main Results:

• The developed matching paradigm demonstrated effective results, with median matching function values ranging from 0.93 to 0.95.
• Soft reconstruction kernels (noise-reducing) from one vendor generally matched with soft kernels from other vendors.
• Sharper reconstruction kernels were similarly matched across different vendors, indicating successful harmonization.

Conclusions:

• A combined quantitative assessment of NPS and MTF provides an effective strategy for harmonizing technical image quality between diverse CT scanners.
• The study successfully matched reconstruction algorithms and kernels, leading to more consistent CT image quality.
• A software tool was shared to aid other institutions in achieving CT image quality harmonization.