PT - JOURNAL ARTICLE AU - Simin Razavi AU - Lionel Zuckier AU - Ran Klein TI - <strong>Optimization of Image Acquisition and Reconstruction Parameters for Quantitative <sup>99m</sup>Tc SPECT-CT in Phantom and Patients</strong> DP - 2017 May 01 TA - Journal of Nuclear Medicine PG - 1365--1365 VI - 58 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/58/supplement_1/1365.short 4100 - http://jnm.snmjournals.org/content/58/supplement_1/1365.full SO - J Nucl Med2017 May 01; 58 AB - 1365Objectives: To optimize factors influencing the accuracy of quantitative 99mTc SPECT-CT using known activity concentration sources in both experimental phantoms and patient acquisitions.Methods: A Jaszczak phantom consisting of six cold spheres and bars was filled with 99mTc-pertechnate of known concentration. In addition, 4 containers of varied sizes were also filled with the same activity concentration as the phantom. All containers were positioned in a SPECT-CT scanner. Four projection sets (permutations of continuous versus step-and-shoot and circular versus non-circular orbits) were acquired followed by a single CT. Tomographs were quantitatively reconstructed with HybridRecon 2.1 (Hermes Medical Solutions, Stockholm, Sweden) in Bq/cc units using several available algorithms (OSEM, MAP-Smoothing, AMAP, MAP-MRP, and MAP-MRP-FMH) and different OSEM reconstruction parameters (4 versus 20 iterations and with versus without resolution recovery (RR)). For all reconstructed images, 1.0 cm diameter, spherical volumes of interest (VOIs) were centered in each container and mean activity concentration was sampled in MBq/cc. The ratio of measured to calibrated activity concentration was calculated as the recovery coefficient (RC), with RC=1 assumed ideal. Factors potentially influencing RC including acquisition modes, reconstruction algorithm, OSEM reconstruction parameters and container sizes were examined by ANOVA (at p=0.05). The optimal acquisition and reconstruction methods were then used in clinical studies in which sources (4.2 cm diameter cylinders) of known activity concentration were placed within the axillae and/or between the legs of patients, and RC in a clinical setting was measured. VOIs were drawn 4 times to assess operator variability.Results: Acquisition modes did not significantly affect the RC accuracy; however, reconstruction methods and container sizes did. OSEM parameters (i.e. iterations and RR) did significantly influence mean RC with 4 iterations and RR giving the optimum results (RC=0.89±0.03, p&lt;0.0001). Optimal accuracy (RC=0.97±0.03) was achieved in 2.7 cm diameter cylinders with underestimation in both the smaller and larger containers (p&lt;0.001). MAP reconstructions achieved significantly better mean RC for all container sizes (0.95±0.07, p&lt;0.01 compared to OSEM 4 iterations and RR), with MAP-MRP determined to be the most accurate (RC=0.96±0.01). In clinical data (10 patients with 23 sources) using MAP-MRP reconstruction RC was 1.03±0.04, slightly overestimating activity concentrations (p&lt;0.001). Operator dependent variability was 4.6%.Conclusion: As determined in phantom studies, and validated in patient studies, quantitative SPECT-CT is robust with respect to varying image acquisition methods, however accuracy is impacted by choice of reconstruction algorithms and is also subject to partial volume and edge ringing artifacts. MAP reconstruction using median root priors (MAP-MRP) achieved nearly ideal accuracy in containers with diameters as small as 1.5× image spatial resolution, with other methods slightly underestimating activity concentrations. Research support: This research was kindly supported by Hermes Medical Solutions.