Comparison of automatic and manual quality control procedures for a Siemens SPECT/CT system.

Objectives The quality of Nuclear Medicine scans is dependent on the reproducibility of clinical radionuclide counting and the accuracy of the instrumentation. Quality control (QC) is an established set of corrections and analyses that validate camera performance within a predefined range. Siemens Symbia SPECT/CT systems come equipped with both user-initiated (manual) and automated QC procedures. Using statistical analysis and phantom imaging, the objective of this study was to evaluate potential differences in reconstructed SPECT uniformity, contrast and resolution when comparing automatic and manual QC calibrations.

Methods To quantify the reproducibility of clinical imaging QC, five trials of MHR and uniformity calibrations were acquired both manually and automatically per the Siemens QC protocol recommendations. Manual multi-head registration (MHR) correction data was collected using five 37 MBq point sources of TC-99m in an integrated point source holder. Automatic MHR data was collected using an internal Gd-153 line source within the Siemens SPECT/CT system. All images were attained with LEHR collimators with the time per projection set to acquire 50 k counts on the first acquired projection. Manual intrinsic uniformity was evaluated using a 1 MBq Tc-99m point source placed into an integrated source holder to create a uniform flood. A 200 M count image was obtained. Automatic uniformity was acquired using an internal Co-57 point source to obtain a 200 M count image. Both automatic and manual MHR data, along with the uniformity calibration values, were statistically analyzed for variance and significance. In addition, a Spectrum Deluxe phantom (Jaz) was imaged using the ACR standard protocol for SPECT systems following both the manual and automatic calibration procedures. Visual analysis of reconstructed uniformity, contrast and resolution was performed.

Results The center of rotation data demonstrated a variance range of 0.02 and 0.25 for the manual and automatic MHR calibrations, respectively. There was no statistically significant difference between manual and automatic MHR parameters (p-value > 0.05). When analyzing uniformity values, there was a variance range of 0.001 and 0.009 over all the data, for both the manual and the automatic data. There was no statistically significant difference between manual and automatic uniformity parameters (p-value > 0.8). However, the reconstructed Jaz images demonstrated improved contrast of cold spheres and improved resolution of the cold rods.

Conclusions Although there was no statistically significant difference between the mean values from manual and automatic QC procedures, the SPECT reconstructed images from the phantom did demonstrate statistically significant improvement of contrast and resolution using the automatic QC procedures.