RT Journal Article
SR Electronic
T1 Normal V/Q SPECT-CT model for monte-carlo studies
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 1725
OP 1725
VO 59
IS supplement 1
A1 David Bourhis
A1 Marine Essayan
A1 Philippe ROBIN
A1 Ronan Abgral
A1 Le Duc Pennec Alexandra
A1 Pierre Salaun
A1 Pierre Yves Le Roux
YR 2018
UL http://jnm.snmjournals.org/content/59/supplement_1/1725.abstract
AB 1725Objectives: Monte-carlo (MC) studies have proven their value in various fields of nuclear medicine. If planar V/Q examinations have often been studied with MC, only few data is available on dual isotopes SPECT examinations. The aim of this study was to create a realistic monte-carlo model of normal dual-isotopes lung ventilation and perfusion single photon emission tomography (V/Q-SPECT).Methods: The MC code used for this study was SIMIND program. The first step was to validate the gamma camera modeling by comparing acquired and simulated data. All acquisitions were performed on a Siemens Symbia T6 gamma camera equipped with a medium energy collimator with 99mTc solutions. Various quality phantoms were used. Although the study was about dual isotopes 99mTc-81mKr acquisitions, the validation was made only with 99mTc, as it is impossible to measure 81mKr activity. Spatial resolution was measured and simulated with a 21.3MBq 99mTc point source. Planar acquisitions (0.12mm pixels) were performed and simulated from 5 to 25 cm between source and detector, and the full width half maximum (FWHM) was calculated as a function of the distance. Root Mean Square Error (RMSE) was then calculated. Planar sensibility was measured and simulated with a 65.23Mbq 99mTc filled phantom at 10cm from detector and relative difference was calculated. Activity recovery was measured and simulated with a NEMA-NU2 2007 phantom. Hot spheres were filled with a 486 kBq.mL-1 99mTc solution and background with a 45.6 kBq.mL-1 solution, which led to a 9.4 concentration ratio. SPECT-CT acquisition was performed and simulated (128² matrix, 128 projections, OSEM3D reconstruction with collimator blurring, scatter and attenuation correction), activity recovery as function of sphere volume was measured and RMSE was calculated. The second step was to create a voxelized phantom for the simulation. A real normal CT scan was used and segmented in 5 density zones (air, lungs, fat, flesh, bones). 2 different activity concentration maps were used: a homogeneous map for ventilation SPECT, and a 10 horizontal parallel zones map to model the anterior-posterior activity concentration gradient for perfusion SPECT. This gradient was modeled thanks to a measurement on 10 normal lung perfusion SPECT. The third step aimed to validate the model by comparing the simulated model to a normal cases database, built with 76 co-registered normal VQ/SPECT-CT. After a non-rigid registration with MiM software (MiM®, version 6.7.8) between the database and the simulation, voxelized Z scores were calculated and expressed as median [minimum, Q1, Q3, maximum]. Results: Concerning the gamma camera model validation, the RMSE between the spatial resolution measure and simulation was 1.6mm, the relative difference between the measured and simulated sensitivity was 4.6%, the RMSE between the measured and simulated activity recovery was 0.86%. The anterior-posterior gradient model was a 2nd order polynomial function which fitted the average gradient on normal cases with R²=0.999. Concerning the realistic model simulation, compared to the normal cases database, Z-score were -0.2 [-3.1, -0.8, 0.6, 6.7] for 81mKr ventilation and 0 [-2.6, -0.6, 0.5, 4.1] for 99mTc perfusion respectively. Conclusion: The SIMIND MC package enables the simulation of a realistic normal dual isotopes lung V/Q-SPECT. Compared to a normal cases database, most simulated voxels have Z-scores close to 0. This model may be used to create pulmonary embolism models and then to develop automated contouring methods for regional lung dysfunction measurements.