Practical issues and limitations of brain attenuation correction on a simultaneous PET-MR scanner

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Radiology

Background:

• Clinical PET-MR imaging, despite its routine use since 2011, faces challenges in attenuation correction.
• Issues affecting brain PET-MR accuracy include CT-to-PET conversion, MR head coil padding, hair, and headphones.

Purpose of the Study:

• To investigate the impact of specific factors on the reliability and accuracy of brain PET-MR attenuation correction.
• To evaluate practical solutions for mitigating identified issues in clinical and research settings.

Main Methods:

• Examined the effect of CT Hounsfield unit conversion to 511 keV linear attenuation coefficients.
• Assessed the influence of MR head coil padding, dense hair, and headphones on reconstructed PET images.
• Evaluated practical mitigating measures for each identified issue.

Main Results:

• CT data conversion is feasible using vendor-specific methods; head coil padding has minimal effect.
• Dense hair significantly impacts reconstructed PET images (>10% error).
• Headphones cause substantial errors if not included in attenuation maps, but scanning without them may be acceptable based on sound levels.

Conclusions:

• Limitations in PET-MR attenuation correction must be considered for accurate brain PET quantification in research and clinical protocols.
• Dense hair, headphones, and independent CT scans can cause non-negligible PET quantification errors.
• Pragmatic approaches can minimize these effects, though they add complexity compared to PET-CT.