The effect of tissue-segmented attenuation maps on PET quantification with a special focus on large arteries

Accuracy on quantitative PET image analysis relies on the correct application of attenuation correction which is one of the major challenges for PET/MRI that remains to be solved. The purpose of this study is to evaluate the effect of MRI-based attenuation maps and the use of flexible coils on the quantitative accuracy of PET images with a special focus on large arteries.

PET/CT data from eight oncologic patients was used. PET data was reconstructed using attenuation maps with different level of detail emulating several approaches available on current PET/MRI scanners. PET images obtained with CT-based and MRI-based attenuation maps were compared to evaluate the quantitative biases obtained. The quantitative effect produced by flexible MRI receiver coils on the attenuation maps was also studied.

The use of simpler attenuation maps produced increased biases between PET data reconstructed with CT-based and MRI-based attenuation maps for fat, non-fat soft-tissues and bone. Biases in lung were very high due to the large heterogeneity and inter-patient variability of the lung. The quantification on large arteries had small deviations except for the case when flexible coils were used. The TBR provided smaller biases in all cases as it cancelled out the similar deviations obtained for arteries and reference veins.

Simplified attenuation maps used on PET/MRI significantly increase the quantitative variability of PET images especially on lungs and bones. The quantification of PET images acquired with PET/MRI scanners applied to studies of atherosclerosis has small deviations, especially when the TBR is considered.