Razvan Ionasec PhD, Presenter: Employee, Siemens AG

Bogdan Georgescu PhD, Abstract Co-Author: Employee, Siemens AG

Dorin Comaniciu, Abstract Co-Author: Employee, Siemens AG

Sebastian Vogt, Abstract Co-Author: Employee, Siemens AG

U. Joseph Schoepf MD, Abstract Co-Author: Speakers Bureau, Bracco Group Speakers Bureau, General Electric Company Speakers Bureau, Bayer AG Speakers Bureau, TeraRecon, Inc Medical Advisory Board, Bracco Group Medical Advisory Board, General Electric Company Medical Advisory Board, Bayer AG Research grant, Bayer AG Research grant, Bracco Group Research grant, General Electric Company Research grant, Bayer AG Research grant, Siemens AG

Eva Maria Gassner MD, Abstract Co-Author: Nothing to Disclose

Quantitative and visual evaluation of the aortic root from CT derived multiplanar images is potentially affected by approximation errors and measurement inaccuracies. Volumetric display provides comprehensive morphologic and dynamic information, however it is difficult to obtain precise measurements of root dimensions or leaflets dynamics. We introduce a 4D model for quantitative evaluation of the aortic root, based on patient specific volumetric data sets.

A model of the aortic root is created by using landmarks, parametric geometries and topological constraints for fast tracking of anatomic structures. The model provides a four-dimensional representation of the hinges, commissures, ostia, leaflets and aortic root surfaces. Discriminative learning-algorithms are applied to facilitate automated model fitting to individual volumetric images from cardiac MDCT. We investigated the accuracy of our method in modeling the aortic root and the automated fitting performance by computing the detection error at defined landmarks for 37 cardiac CT data sets. Additionally, in four exemplary data sets (unaffected root, stenotic valve, dilated root and bicuspid valve), model-driven quantification of root dimensions and valve dynamics was compared to manually obtained parameters from CT images.

Evaluation of automated model-fitting demonstrated a mean detection error of 1.33mm. Model-driven quantification of 4D geometries could reproduce typical motion patterns of the aortic root and leaflets known from experimental and echocardiographic studies, such as abnormal dynamics for bicuspid and restricted patterns for stenotic valves. The leaflets of a bicuspid prolapsing valve showed a hypermobile excursion of 21.5mm, compared to functional valves with a motion range of 8.1mm. Measurements of stenotic valves demonstrated restricted opening patterns, with an excursion range of 6.7mm and delayed slope. The measurements accuracy revealed mean errors of 0.21cm and 0.13cm for root diameters and leaflet motion, respectively.

Combined visual evaluation and reliable quantification are enabled by a 4D aortic root model. Results demonstrate the potential in performing standard measurements as well as in allowing quantitative assessment of complex geometries and dynamics.

Automated assessment of aortic root morphology and function may improve the consistency of aortic valve evaluation at MDCT.

Ionasec, R, Georgescu, B, Comaniciu, D, Vogt, S, Schoepf, U, Gassner, E, Patient-specific 4D Aortic Root Models Derived from Volumetric Image Data Sets.  Radiological Society of North America 2008 Scientific Assembly and Annual Meeting, February 18 - February 20, 2008 ,Chicago IL. http://archive.rsna.org/2008/6022710.html