Intersoftware variability in SPECT quality control: A technical note on analytical discrepancies and compliance...

Area of Science:

• Medical Imaging Physics
• Diagnostic Radiology
• Quality Assurance in Healthcare

Background:

• Single-Photon Emission Computed Tomography (SPECT) systems require rigorous quality control (QC) for reliable diagnostic imaging.
• Current reliance on diverse third-party software for SPECT QC analysis introduces non-standardized variability.
• This variability can compromise accurate equipment performance assessment and interinstitutional comparability.

Purpose of the Study:

• To evaluate the extent of variability in SPECT quality control (QC) test results arising from different image analysis software packages.
• To highlight the critical need for standardization in SPECT QC software to ensure consistent and reliable performance metrics.

Main Methods:

• Five commercial SPECT QC software packages (A-E) were utilized for analysis.
• Identical DICOM image datasets from four SPECT/CT systems were analyzed following the WS 523-2019 standard.
• Key performance indicators including file reading success, intrinsic uniformity, resolution, linearity, and compliance rates were assessed using statistical methods like ANOVA and Welch's tests.

Main Results:

• Significant variations were observed in file reading success rates (61.8%-100%) and compliance rates (68.8%-100%) across software packages.
• Statistically significant intersoftware differences were found for intrinsic integral uniformity, spatial resolution, and differential linearity.
• Large effect sizes (η² > 0.14) for spatial resolution and differential linearity indicate substantial software-driven discrepancies.

Conclusions:

• Significant disparities exist between SPECT QC software packages in file reading and performance indicator analysis.
• These software-induced differences directly affect the accuracy of equipment evaluations and interinstitutional comparability.
• Divergent conclusions on device compliance status can arise due to the use of different analysis software.