Slice NEQ and system DQE to assess CT imaging performance

Area of Science:

• Medical Imaging Physics
• Radiological Sciences
• Image Quality Assessment

Background:

• Standard Fourier metrics for CT image quality, like 2D Noise Equivalent Quanta (NEQ), do not account for longitudinal resolution variations.
• This omission introduces bias when comparing image quality across different reconstructed slice thicknesses.
• Accurate volumetric assessment of scanner performance requires metrics that integrate all spatial dimensions.

Purpose of the Study:

• To develop and validate a "slice NEQ" metric for volumetric assessment of CT scanner imaging performance.
• To introduce a system Detective Quantum Efficiency (DQE) formulation specific to the CT isocenter.
• To address the limitations of existing 2D metrics in evaluating image quality with varying slice thicknesses.

Main Methods:

• Designed a "slice NEQ" metric incorporating longitudinal resolution for volumetric image quality assessment.
• Developed a system DQE formulation for the CT isocenter.
• Utilized a water phantom with contrast inserts and a Catphan® 600 phantom for task-based transfer function (TTF) and noise power spectra (NPS) measurements.
• Acquired and reconstructed images on GE and Siemens CT scanners with varying slice thicknesses (1-5 mm) and reconstruction algorithms (FBP, ASIR).

Main Results:

• The "slice NEQ" accurately compared imaging performance across different longitudinal resolutions.
• Standard 2D NEQ values increased with reconstructed slice thickness, indicating a limitation.
• System DQE peaked at 0.70 for Siemens and 0.50 for GE (FBP reconstructions).
• Fourier-based metric validity was limited with iterative reconstruction due to nonlinearities.

Conclusions:

• The proposed "slice NEQ" offers a more accurate volumetric assessment of CT scanner performance than traditional 2D NEQ.
• The developed metrics provide improved comparability of image quality across varying slice thicknesses and scanner configurations.
• Further research is needed to address nonlinearities in iterative reconstruction for comprehensive metric application.