Three-dimensional morphometry of the human thoracic aorta using centerline analysis based on least-squares plane fitting

Abstract

ObjectiveA novel approach to 3-dimensional morphometry of the thoracic aorta was developed by applying centerline analysis based on least-squares plane fitting, and a preliminary study was conducted using computed tomography imaging data. MethodsWe retrospectively compared 3 groups of patients (16 controls without aortic disease, and 16 cases each with acute type B aortic dissection and congenital bicuspid aortic valve). In addition to the standard assessment indices for curvature κ and torsion τ, we conducted coordinate transformation based on the least-squares plane, divided the centerline into 3 representative features (transverse, anterior-posterior, and longitudinal displacements), and analyzed the overall and local displacement in each direction. The transverse displacement, represented by the distance of the centerline from the least-squares plane, was curve-fitted to the damped oscillation waveform. Thereafter, damped oscillation parameters were compared for each group. ResultsCurvature κ exhibited a bimodal distribution, with peaks observed in the ascending aorta and aortic arch, and torsion τ exhibited a transition from positive to negative values in the arch. There were significant differences in the mean displacement between the groups for each direction (transverse P = .0083, anteroposterior P = .010, longitudinal P = 1.32 × 10−6). Furthermore, interval integral analysis revealed that several intervals exhibited significant differences between groups in each direction. The amplitude of damped oscillation parameters was significantly larger in the bicuspid aortic valve group than in the control and type B aortic dissection groups. ConclusionsThe novel analytical approach permitted a quantitative assessment of the 3-dimensional morphological differences between the control, type B aortic dissection, and bicuspid aortic valve groups.