Quantification and homogenization of image noise between two CT scanner models

Area of Science:

• Medical Imaging
• Radiology
• Image Quality Assessment

Background:

• Radiologists reported image appearance and noise differences between new and existing CT scanners.
• These discrepancies, despite identical reconstruction algorithms, hindered post-contrast CT scan interpretation.
• Standardized image quality is crucial for consistent diagnostic performance.

Purpose of the Study:

• Quantify image quality differences between a new and existing CT scanner.
• Investigate scanner-specific variations in CT number and noise magnitude.
• Standardize image quality across different CT scanners from the same vendor.

Main Methods:

• Analyzed daily quality control images (water phantom) for CT number and noise magnitude.
• Generated noise power spectra (NPS) using American College of Radiology CT phantom uniformity sections.
• Imaged liver and multi-energy phantoms with varying material attenuations on both scanners.
• Utilized statistical tests to determine significant differences between scanners.

Main Results:

• The new scanner showed slightly higher CT number and lower noise magnitude in daily QC images.
• No significant difference in measured noise power spectra was found between scanners.
• New scanner's noise magnitude increased with material attenuation, unlike the existing scanner.
• An alternate reconstruction algorithm on the new scanner resolved high attenuation deviations.

Conclusions:

• Standard water phantom noise evaluation was insufficient to detect scanner-specific image quality differences.
• Comprehensive phantom testing, including high attenuation materials, is essential for characterizing and standardizing CT image quality.
• Image quality homogenization between CT scanners is achievable through advanced characterization and algorithm adjustment.