Dixon imaging-based partial volume correction improves quantification of choline detected by breast 3D-MRSI.

Objectives: Our aim was to develop a partial volume (PV) correction method of choline (Cho) signals detected by breast 3D-magnetic resonance spectroscopic imaging (3D-MRSI), using information from water/fat-Dixon MRI. Methods: Following institutional review board approval, five breast cancer patients were measured at 3 T. 3D-MRSI (1 cm resolution, duration ~11 min) and Dixon MRI (1 mm, ~2 min) were measured in vivo and in phantoms. Glandular/lesion tissue was segmented from water/fat-Dixon MRI and transformed to match the resolution of 3D-MRSI. The resulting PV values were used to correct Cho signals. Our method was validated on a two-compartment phantom (choline/water and oil). PV values were correlated with the spectroscopic water signal. Cho signal variability, caused by partial-water/fat content, was tested in 3D-MRSI voxels located in/near malignant lesions. Results: Phantom measurements showed good correlation ( r = 0.99) with quantified 3D-MRSI water signals, and better homogeneity after correction. The dependence of the quantified Cho signal on the water/fat voxel composition was significantly ( p < 0.05) reduced using Dixon MRI-based PV correction, compared to the original uncorrected data (1.60-fold to 3.12-fold) in patients. Conclusions: The proposed method allows quantification of the Cho signal in glandular/lesion tissue independent of water/fat composition in breast 3D-MRSI. This can improve the reproducibility of breast 3D-MRSI, particularly important for therapy monitoring. Key Points: • Choline signal in breast spectroscopic imaging can be corrected using Dixon MRI. • Partial volume-matched MRI provides comparable information with quantified 3D-MRSI water signal. • Corrected breast 3D-MRSI could improve reproducibility for follow-up studies and inter-subject comparison. [ABSTRACT FROM AUTHOR]