Quantitative Analysis of Geometry and Lateral Symmetry of Proximal Middle Cerebral Artery

Background: The purpose of our work is to quantitatively assess clinically relevant geometric properties of proximal middle cerebral arteries (pMCA), to investigate the degree of their lateral symmetry, and to evaluate whether the pMCA can be modeled by using state-of-the-art deformable image registration of the ipsi- and contralateral hemispheres. Methods: Individual pMCA segments were identified, quantified, and statistically evaluated on a set of 55 publicly available magnetic resonance angiography time-of-flight images. Rigid and deformable image registrations were used for geometric alignment of the ipsi-and contralateral hemispheres. Lateral symmetry of relevant geometric properties was evaluated before and after the image registration. Results: No significant lateral differences regarding tortuosity and diameters of contralateral M1 segments of pMCA were identified. Regarding the length of M1 segment, 44% of all subjects could be considered laterally symmetrical. Dominant M2 segment was identified in 30% of men and 9% of women in both brain hemispheres. Deformable image registration performed significantly better (P < .01) than rigid registration with regard to distances between the ipsi-and the contralateral centerlines of M1 segments (1.5 +/- 1.1 mm versus 2.8 +/- 1.2mm respectively) and between the M1 and the anterior cerebral artery (ACA) branching points (1.6 +/- 1.4 mm after deformable registration). Conclusions: Although natural lateral variation of the length of M1 may not allow for sufficient modeling of the complete pMCA, deformable image registration of the contralateral brain hemisphere to the ipsilateral hemisphere is feasible for localization of ACA-M1 branching point and for modeling 71 +/- 23% of M1 segment.