Attenuation Correction Approaches for Serotonin Transporter Quantification With PET/MRI.

Area of Science:

• Neuroimaging
• Radiochemistry
• Medical Physics

Background:

• Accurate attenuation correction (AC) is critical for hybrid PET/MR imaging, especially for absolute quantification in neurotransmitter system studies.
• Existing MR-based AC methods are often evaluated using standardized uptake values, but absolute quantification is preferred for accuracy.
• This study investigates the accuracy of various MR-based AC methods for quantifying serotonin transporter (SERT) parameters.

Purpose of the Study:

• To evaluate the accuracy of segmentation- and atlas-based MR AC approaches in hybrid PET/MR imaging.
• To assess the impact of different AC methods on the quantification of serotonin transporter (SERT) distribution volumes and occupancy.
• To compare MR-based AC methods against a computed tomography (CT) reference standard.

Main Methods:

• Eighteen healthy subjects underwent [11C]DASB PET/MRI scans with either citalopram or placebo infusion.
• Six AC methods were evaluated: DIXON, UTE, pseudoCT (derived from T1-weighted MRI), bone-enhanced DIXON, and RESOLUTE (optimized UTE segmentation).
• Quantitative parameters including total and specific distribution volume (V_T, V_S = BP_P) and occupancy were calculated and compared to CT-based AC.

Main Results:

• All AC methods showed bias in V_T and BP_P quantification, ranging from -15% to 17%.
• The pseudoCT approach yielded the smallest relative errors (<5%) and best agreement with CT.
• While DIXON and UTE showed significant bias in differentiating BP_P between SSRI and placebo conditions, they maintained high correlation with CT (r²~1); occupancy errors were small (<4%) in high-binding SERT regions across all methods.

Conclusions:

• The pseudoCT method offers the most accurate attenuation correction for SERT quantification in PET/MRI.
• High correlations observed with CT suggest that biased V_T and BP_P values may be rescaled to true values.
• For SERT occupancy, all tested AC methods demonstrated minimal bias in high-binding regions, suggesting potential for reliable longitudinal studies.