Identifying the principal modes of variation in human thoracic aorta morphology

J Thorac Imaging. 2014 Jul;29(4):224-32. doi: 10.1097/RTI.0000000000000060.


Purpose: Diagnosis and management of thoracic aorta (TA) disease demand the assessment of accurate quantitative information of the aortic anatomy. We investigated the principal modes of variation in aortic 3-dimensional geometry paying particular attention to the curvilinear portion.

Materials and methods: Images were obtained from extended noncontrast multislice computed tomography scans, originally intended for coronary calcium assessment. The ascending, arch, and descending aortas of 500 asymptomatic patients (57 ± 9 y, 81% male) were segmented using a semiautomated algorithm that sequentially inscribed circles inside the vessel cross-section. Axial planes were used for the descending aorta, whereas oblique reconstructions through a toroid path were required for the arch. Vessel centerline coordinates and the corresponding diameter values were obtained. Twelve size and shape geometric parameters were calculated to perform a principal component analysis.

Results: Statistics revealed that the geometric variability of the TA was successfully explained using 3 factors that account for ∼80% of total variability. Averaged aortas were reconstructed varying each factor in 5 intervals. Analyzing the parameter loadings for each principal component, the dominant contributors were interpreted as vessel size (46%), arch unfolding (22%), and arch symmetry (12%). Variables such as age, body size, and risk factors did not substantially modify the correlation coefficients, although some particular differences were observed with sex.

Conclusions: We conclude that vessel size, arch unfolding, and symmetry form the basis for characterizing the variability of TA morphology. The numerical data provided in this study as supplementary material can be exploited to accurately reconstruct the curvilinear shape of normal TAs.

MeSH terms

  • Adult
  • Aged
  • Aged, 80 and over
  • Algorithms
  • Aorta, Thoracic / diagnostic imaging
  • Aorta, Thoracic / pathology*
  • Aortic Diseases / diagnostic imaging
  • Aortic Diseases / pathology*
  • Body Surface Area
  • Female
  • Humans
  • Image Processing, Computer-Assisted
  • Male
  • Middle Aged
  • Multidetector Computed Tomography
  • Principal Component Analysis
  • Retrospective Studies