PT - JOURNAL ARTICLE
AU - Moore, Bria
AU - Galt, James
AU - Mirando, David
AU - Nelson, Aaron
AU - Nye, Jonathon
TI - Phantom evaluation of a vendor neutral quantitative SPECT/CT reconstruction package
DP - 2019 May 01
TA - Journal of Nuclear Medicine
PG - 1359--1359
VI - 60
IP - supplement 1
4099 - http://jnm.snmjournals.org/content/60/supplement_1/1359.short
4100 - http://jnm.snmjournals.org/content/60/supplement_1/1359.full
SO - J Nucl Med2019 May 01; 60
AB - 1359Introduction: Advances in quantitative SPECT/CT imaging have found clinical utility with the calculation of standardized uptake values and evaluation of internal dosimetry. MIM SPECTRA Quant (MIM Software Inc.) provides a vendor neutral quantitative SPECT/CT reconstruction using an ordered subset expectation maximization (OSEM), with CT-based attenuation correction, dual or triple energy window scatter correction and resolution recovery. This study aimed to compare the quantitative accuracy of MIM SPECTRA Quant package and Siemens xSPECT on the Siemens Symbia Intevo T16 SPECT/CT platform. Methods: A standard NEMA IEC body phantom with 6 fillable spheres (10, 13, 17, 22, 28 and 37 mm dia.) was filled with a concentration ratio of about 9:1 (sphere: background). The camera was calibrated using two approaches 1) a 6.3 L uniform right cylinder filled to a concentration of 0.42 μCi/mL and 2) a 3.23 mCi point source of Tc-99m placed in the central field-of-view. Data were collected on the NEMA and uniform phantoms using the following protocol: 120 total projections, 25 sec/proj, noncircular orbit, 15% photopeak window, and 15% lower scatter window. Three reconstruction methods were employed, all including 3D resolution recovery, CT-based attenuation correction and dual window scatter correction: A) Siemens xSPECT, an ordered subset conjugation gradient reconstruction (OSCG), B) Siemens FLASH 3D, an OSEM reconstruction, and C) MIM SPECTRA Quant, an OSEM reconstruction. Reconstruction parameters were set to 10 subsets while varying the number of iterations from 4 to 10 all with a 6mm Gaussian post-filter. The uniform phantom calibration was compared to the xSPECT sensitivity factor calculated from the planar point source acquisition. Because the Flash3D projection data are scaled during reconstruction it was not possible to derive calibration factors from a separate image for the NEMA phantom. Instead, the NEMA image data were calibrated to the mean value of a large volume of interest (VOI) placed in a uniform section of known activity concentration. VOIs were drawn on each sphere that matched their inner diameter on the CT image including a large background VOI located in a uniform section. Recovery coefficients were calculating by dividing the mean sphere VOI activity concentration by the truth. Results: The calibration factor derived from the uniform phantom and planar point source acquisitions varied by less than a percent (0.53). We found very little variation in calculated SUV values with increasing number of iterations but small improvements in the recovery coefficient (RC) were observed for spheres smaller than 28 mm for all reconstructions. Focusing on the group of images reconstructed with 8 iterations and 10 subsets, the maximum mean recovery coefficient in the largest sphere was 0.9 using xSPECT, 0.88 using FLASH3D, and 0.85 for MIM SPECTRA Quant. Conclusions: This study determined that the MIM SPECTRA Quant software is capable of SPECT quantitative imaging and compares well with xSPECT and Flash3D reconstructions on a Symbia Series SPECT/CT. We are continuing work to optimize the attenuation and scatter models to further improve MIM SPECTRA Quant’s accuracy.