Vessel segmentation for χ $$ \chi $$ -separation in quantitative susceptibility mapping.

Area of Science:

• Neuroimaging
• Medical Physics
• Biomedical Engineering

Background:

• Quantitative susceptibility mapping (QSM) generates paramagnetic (χpara) and diamagnetic (|χdia|) maps, reflecting brain iron and myelin distribution.
• Vessel artifacts in QSM interfere with accurate quantification of iron and myelin.
• Advanced QSM methods like χ-separation aim to improve tissue characterization.

Purpose of the Study:

• To develop a novel vessel segmentation method for χ-separation to address artifacts caused by vessels.
• To improve the accuracy of iron and myelin quantification in QSM applications.

Main Methods:

• A three-step method involving seed generation, guided region growing, and mask refinement.
• Utilized R2* and the product of χpara and |χdia| maps for seed generation.
• Compared performance against existing vessel segmentation techniques qualitatively and quantitatively.

Main Results:

• The proposed method demonstrated superior performance, achieving high Dice scores (e.g., 76.7% for χpara at 3T).
• Effectively excluded non-vessel structures, improving accuracy in QSM analysis.
• Showed notable improvements in quantitative evaluation of χ-sepnet-R2* and significant differences in ROI analysis.

Conclusions:

• The developed vessel segmentation method generates high-quality vessel masks for QSM.
• This technique has the potential to facilitate various QSM applications by enabling more reliable analysis.