TY - JOUR T1 - Development and implementation of a Joint-Commission compliant phantom program for PET/CT with fully-automated cloud-based analysis. JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 570 LP - 570 VL - 59 IS - supplement 1 AU - John Sunderland AU - Christian Bauer AU - Reinhard Beichel AU - Ethan Ulrich AU - Paul Christian AU - Raymond Muzic AU - Martin Lodge AU - Jonathon Nye AU - Lance Burrell AU - Michael Czachowski AU - Patrick Wojtylak AU - Leonard Zimmermann Y1 - 2018/05/01 UR - http://jnm.snmjournals.org/content/59/supplement_1/570.abstract N2 - 570Objectives: In 2015 The Joint Commission published new “Diagnostic Imaging Requirements” that mandated phantom-based PET/CT scanner performance evaluations by diagnostic medical physicists. Image uniformity, high contrast resolution/system spatial resolution, low contrast resolution or detectability, and artifact evaluation are the newly required components. No guidance is given as to which phantoms or what criteria are to be used in these assessments. Since most PET/CT scanners are hospital-based in the US, and 82% of hospitals are currently Joint Commission accredited, these new required evaluations impact the majority of PET/CT scanners in the US. Here we report the development of a phantom-based PET/CT scanner evaluation program designed to be compliant with Joint Commission requirements while providing meaningful and actionable scanner performance information. Methods: The proposed phantom program requires sites to perform two phantom scans. The first scan images a standard uniform 20 cm diameter cylinder phantom filled with aqueous [18F]FDG. The cylinder is centered in the gantry, but tilted at a slightly oblique angle with respect to the z-axis and is imaged for two bed positions at 20-30 minutes per bed position. Images reconstructed using the site’s standard clinical reconstruction yields data for in-plane uniformity, axial uniformity, assessment of quantitative calibration, and reconstructed resolution in the radial and axial directions. Spatial resolution is calculated using measurement of the edge response function using the method of Lodge and Leal (J Nucl Med 2017, Vol 58 701). The second phantom scan uses either the NEMA Image Quality (IQ) phantom, or the new version of the SNMMI Clinical Trials Network (CTN) oncology phantom with NEMA sized spheres. A 4:1 sphere-to-background ratio is used in either case. The phantom data are acquired using the site’s standard oncology protocol including time per bed position and reconstruction parameters (advanced reconstructions allowed). Image data yields a contrast recovery coefficient curve, clinically relevant noise assessment, and assessment of low-contrast lesion detectability. The two phantom image sets are uploaded to a cloud-based server along with phantom-fill documentation. After a manual image quality control check, the datasets sets are exported to a folder that is continuously interrogated (Python). Upon dataset detection, images are automatically downloaded and analyzed in a fully automated fashion (C++, ITK) and a full scanner performance report (PDF) is generated. The report is designed to demonstrate Joint Commission compliance and record other performance metrics. In the final step, a medical physicist reviews the images and PDF report, performs an assessment for lesion detectability and artifacts, and approves the report. In the current plan, reports will be sent directly to the imaging site or independent contract physicist who requested the analysis for co-signature. Results: Robustness of the automated phantom analysis software for the CTN and NEMA IQ phantoms has been validated across a number of PET/CT systems over a range of orientations and statistical image quality (Ulrich, Med Phys 23 Nov 2017). The uniform phantom algorithm for spatial resolution has been similarly validated. Analysis of three phantom datasets (uniform, NEMA IQ, and CTN) is benchmarked at 5 minutes 20 seconds. Conclusions: An efficient Joint Commission compliant phantom program for PET/CT has been developed and tested using commonly available phantoms and a robust cloud-based analysis tool. Deployment of the program to the general public is planned for the near future. Funding: U01CA140206, R01CA169072 ER -