The effect of emission-transmission contamination of image quantification in PET

Aim: The quantification property of conventional PET scanners is directly depending on the accuracy of the performed attenuation correction and hence on the attenuation map used. The transmission data obtained using Ge-68 rod sources, if not corrected for, as observed in some clinical cases suffers from the contamination of the emitted events. The magnitude and the effect of this unwanted phenomenon depend on the relationship between the strength of the external sources and the concentration of the activity uptake. The purpose of this study is to evaluate the extent of this unwanted effect through phantom measurements and to stress the need of spillover correction on transmission data. Methods: A number of emission and transmission scans using a 3D Hoffman brain phantom with different activity concentrations (30, 15, 7, 3, 1.5 and <0.1 MBq/l) where performed on an ECAT EXACT (Siemens / CTI Inc.). The emission data were acquired in 3D with conventional clinical parameters (128x128 matrix, 15 min). The transmission data were acquired in 2D using external Ge-68 rod sources (3 x 56 MBq) as provided by the vendor. The phantom was in the same position across all the measurements. The acquisitions were reconstructed using Ramp filter and FBP. The emission scan with the highest activity was corrected for attenuation using the segmented maps obtained from different phantom activity concentrations using the ECAT software (7.2). The transmission map with activity < 0.1 MBq/l was used as reference and the activity concentrations in the corrected images were evaluated using a standardizing digitalized brain atlas. Entire lobes were used as Volumes of Interest (VOIs). Results: The segmented mu-values, corresponding to the VOIs under study decreased by x1, x2, x3, x4, and x5 % with decreasing phantom activity. There was a POSITIVE correlation between the mu-values and activity in the phantom (R=xx; p<YYY). The effect of this inaccuracy on the attenuation corrected emission data when using segmented maps was correspondingly z1, z2, z3, z4 and z5%. Conclusion: Emission contamination of transmission data affects the quantitative property of PET images especially in areas with high activity concentration such as tumor areas. Without proper correction for emission contamination, the higher the activity concentration, the less the measured mu-value and the magnitude of attenuation correction will be. Proper spill over correction has therefore to be incorporated and performed in order to achieve linear correspondence between uptake and measured activity.