O.12 - Determination of the influence of respiratory motion on the SUV accuracy in pulmonary lesions

Abstract

Introduction and Aim: Respiratory motion results in reduction of qualitative and quantitative accuracy in PET imaging of the heart, muscles of the diaphragm, liver, spleen, pancreas and imaging of the lung area. It causes blurring of the images and thus a loss of sensitivity in lesion detection. The aim was to determine the importance of respiratory motion on SUV accuracy in pulmonary lesions through simulation of respiratory motion. Materials and Methods: The XCAT Phantom was used to assess the influence of breathing on SUV quantification in a human-like simulation model. Six spherical lesions were introduced into the lungs of the model, three in the left lung and three in the right lung. Activity concentrations assigned to the lesions were varied to create a range of different lesions: background ratios, thus creating a range of contrasts. The simulations were repeated for various lesion sizes. GATE software was used to performMonte Carlo simulations during breathing. True-, scattered-, randomand total coincidences were extracted from the simulated data and reconstructed using OPL-EM and corrected for attenuation. Results: The % trues-to-total coincidences generally followed a downward trend with a decrease in contrast. Scattered coincidences had a more significant contribution at low contrast and when the lesion was closer to the diaphragm. This was greater in 3D simulations than in 2D simulations. Random coincidences were relatively greater in the apical region than in the other lung regions, particularly at low contrast large lesions. This increased for 3D compared to 2D acquisitions. The fraction of true coincidences-to-total coincidences increased when the lesions became smaller, but the relative amount of scattered events also increased. Conclusion: Breathing plays a significant role when determining the SUV of pulmonary lesions after accounting for the size, location, and the amount of activity within the simulated lesions compared to the background activity.