Evaluation of Attenuation Correction Methods for PET/MR Imaging

Elena Rota Kops , Heba Alrakh , Cláudia Régio Brambilla, Jürgen Scheins, Hans Herzog, N. Jon Shah and Christoph Lerche

6th July 2024

A new study from INM-4 assesses the accuracy of various attenuation correction (AC) methods for PET/MR imaging, focusing specifically on the cerebellum. Sixteen subjects underwent [18F]FDG scans with a 3T MR-BrainPET insert and a whole-head CT scan. Researchers compared three MR-based attenuation maps—Boston-MGH, London-UCL, and Juelich-Tx-template-based—to CT-based attenuation maps.

The study utilised high-resolution SUIT atlas data to ensure accurate cerebellar analysis. Results indicated variability in performance among the MR-based methods. The Boston-MGH method had a relative error of 3.85%, the London-UCL method had 6.00%, and the Juelich-Tx method had a lower relative error of 0.25%. The findings underscore that the choice of attenuation correction method can significantly impact the accuracy of cerebellar PET imaging, which is crucial for studies using the cerebellum as a reference region.

The study also highlighted that the accuracy and precision of MR-based methods vary, with the Juelich-Tx method providing the most reliable results compared to the CT-based standard. These results are important for refining PET/MR imaging protocols and improving the reliability of neuroreceptor studies.

Original publication: Attenuation Correction of the Cerebellum in PET/MR Data

Last Modified: 30.07.2024