PET/CT annual AAPM quality control: Practical implications for technologists.

Objectives: In late 2019 the American Association of Physicists in Medicine (AAPM) Task Group 126 published a report, "PET/CT Acceptance Testing and Quality Assurance". The goal was to "provide a standardized set of acceptance and periodic tests that can be easily implemented in a QA program" rather than following the full National Electrical Manufacturers Association (NEMA) NU 2 acceptance standards which can be "challenging requiring specialized phantoms". The purpose of this study was to assess the practical implications for nuclear medicine technologists (NMT) performing annual quality control (QC) tests.

Methods: The five QC tests are: Spatial Resolution & PET/CT Registration (RR), Sensitivity (S), Count Rate Performance (CRP), Image Contrast & Scatter/Attenuation Correction (CSAC), and Image Uniformity (U). RR was tested by scanning capillary tubes filled with both F-18 and CT contrast to determine PET resolution and PET/CT registration accuracy according to the recent NEMA-2018 standards. Three tests (S, CRP, and U) were all performed by scanning a standard water phantom. The S test was performed using AAPM option 2, to only monitor the vendor-specific calibration factor, a simpler alternative to option 1 which requires a NEMA phantom. For the CRP test, a baseline was obtained by scanning the water phantom once each hour for 11 hours. Then three 15 minute follow-up scans were acquired as per AAPM guidelines. The U test was also performed by scanning the water phantom as per protocol "to approximate a standard patient dose". Lastly, CSAC was tested by scanning the American College of Radiology (ACR) PET phantom according to the protocol. Each test was assessed for ease-of-performance and ease-of-data analysis as well as the times required for both performance and analysis. Both ease-of-performance and data analysis were classified as either "routine", "moderate", or "complex" according to specific criteria. The classification of "routine" for ease-of-performance was whether the NMT had previous experience scanning the specific phantom. A classification of "routine" for data analysis meant the analysis was totally automated. For both performance and data analyses, a "moderate" classification was given if the test was new to the NMT and/or manually performed. A "complex" classification was assigned if the performance and analysis were both new, involved tedious steps, and/or required assistance from a physicist. Each test was performed twice and the results were averaged.

Results: Three of five tests were classified as "routine" for ease-of-performance (see table). The CSAC test (using the ACR phantom) was "moderate". Although the RR test was "complex", it was performed by a NMT using the manufacturer’s NEMA acquisition protocol. Ease-of-data analyses for two tests (ACR phantom and U) were classified as "routine" since the SNMMI’s fully automated "phantom analysis toolkit" was used. Both the S and CRP data analyses were classified as "moderate", as the NMT had to manually create spreadsheets to calculate results. As expected, the RR analysis was "complex" with PET resolution determined by a NMT using the manufacturer’s software; however, PET/CT registration analysis required the assistance of a physicist using a different imaging display system. The summed total times for both performance and data analyses by a NMT took longer than the estimated times of AAPM by factors of 1.89, 1.09, 1.25, 1.33, and 2.92, for the RR, S, CRP, CSAC and U tests, respectively.

Conclusions: Although all times took longer than AAPM estimates, they could decrease with more experience. All tests were performed by a NMT except for the PET/CT registration analysis. Nonetheless, if software was available for this registration analysis, then NMTs could perform it. In summary, it is feasible that all five QC scans and four of the five data analyses methods recommended by the AAPM can commonly be performed by a nuclear medicine technologist.