Abstract
1354
Objectives: Attenuation correction in simultaneous PET/MRI was initially implemented using a two-point DIXON MR technique which neglected the bones. More recent implementations involve either using an ultra-short echo time based MR sequences (UTE or ZTE), a bone atlas or novel machine learning approaches which allows for inclusion of bones. Evaluation of the impact on the quantitative accuracy of PET uptake measurement of these various attenuation correction methods is much needed and this abstract presents the development of a lesion insertion tool to evaluate the accuracy of PET/MRI lesion SUV values resulting from the choice of attenuation correction strategy.
Methods: A lesion insertion tool has been developed to insert a spherical lesion into human body habitus from PET/MR images and simulate lesion sinogram data based on desired lesion size, location, activity and scan duration and accounting for attenuation, deadtime, scatter, normalization and Poisson statistics. Simulated sinograms are then added to patient sinograms and images are reconstructed with standard iterative image algorithm. The tool was evaluated using physical phantoms and FDG PET cancer patients imaged by PET/MRI. For those patients, a prior CT was registered to the MR based attenuation correction to provide a reference CT based attenuation correction.
Results: The lesion insertion tool was tested using the NEMA-IEC phantom equipped with its standard set of 6 spheres (10 to 37 mm) filled with a 10:1 ratio with a total activity of 0.55 mCi and scanned for 30min. Simulated lesions of the same diameter were inserted and the contrast recovery coefficients (CRC) of the simulated lesion was compared to the measured values where an excellent agreement was found (Fig.1, left). In the same phantom, a 20mm lesion was placed at various position and images were generated with attenuation coefficient where the lung insert was assigned the air (in red) or bone (in blue) attenuation value. Lesion of 20mm were inserted at various position the pelvis area of cervical cancer patients injected with 18F-FDG and data acquired in PET/MR (fig.2). The simulated lesion data are shown to depict reliably the spatial information observed in patient data when comparing CT_AC reconstruction to MR_AC reconstruction, (DIXON attenuation correction is shown).
Conclusions: Excellent agreement of the inserted spherical lesions to the measured lesions CRC has been achieved and indicate that the proposed method can effectively be used to predict and study the quantitative bias on lesion uptake in patient data resulting from the choice of attenuation correction strategy. The evaluation of the methodology will be presented for the two major manufacturers of clinical PET/MRI cameras.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.