UK audit of single photon emission computed tomography reconstruction software using software generated phantoms

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• Quantitative accuracy in single photon emission computed tomography (SPECT) is crucial for reliable clinical diagnosis.
• Standardized phantom data are essential for evaluating the performance of SPECT reconstruction software.

Purpose of the Study:

• To audit the quantitative performance of SPECT software using standardized phantom data.
• To assess variations in quantitative accuracy across different software packages and revisions.

Main Methods:

• Utilized an analytically generated software phantom with defined structures (cylinder, rods, rings) and a measured line source.
• Distributed projection data to multiple UK centers for reconstruction using a 'ramp only' algorithm.
• Collected quantitative parameters such as mean/standard deviation, maximum counts, and full width at half maximum.

Main Results:

• 115 systems from 100 UK centers participated, representing 12 software providers.
• Significant differences in quantitative parameters were found based on input count densities.
• Substantial variations were noted between different software suppliers and even between revisions from the same supplier.

Conclusions:

• Current SPECT software exhibits considerable variability in quantitative accuracy.
• Software performance is influenced by input data characteristics and specific vendor implementations.
• Further standardization and quality control are necessary to ensure consistent and reliable SPECT imaging outcomes.