Subject-specific finite-element modeling of normal aortic valve biomechanics from 3D+t TEE images

• Published in: Medical image analysis 2015-02, Vol.20 (1), p.162-172
• Main Authors: Labrosse, Michel R., Beller, Carsten J., Boodhwani, Munir, Hudson, Christopher, Sohmer, Benjamin
• Format: Article
• Language: English
• Subjects: 3D+t ultrasound imaging; Aged; Aortic Valve - physiology; Biomechanical Phenomena; Computer Simulation; Echocardiography, Three-Dimensional - methods; Finite Element Analysis; Finite element biomechanical models; Humans; Middle Aged; Models, Cardiovascular; Normal aortic valve; Observer Variation; Subject-specific anatomy
• ISSN: 1361-8415; 1361-8423
• DOI: 10.1016/j.media.2014.11.003

[Display omitted] •We propose an integrated framework to process subject-specific 3D+t TEE aortic valve (AV) data.•We acquired the 3-D coordinates of 21 anatomical landmarks of the AV apparatus in 10 datasets.•We evaluated the measurement accuracy of the proposed methodology.•We determined age-matched material properties for the aortic and leaflet tissues.•We automatically built 10 finite element models and simulated the AV function. In the past decades, developments in transesophageal echocardiography (TEE) have opened new horizons in reconstructive surgery of the aortic valve (AV), whereby corrections are made to normalize the geometry and function of the valve, and effectively treat leaks. To the best of our knowledge, we propose the first integrated framework to process subject-specific 3D+t TEE AV data, determine age-matched material properties for the aortic and leaflet tissues, build a finite element model of the unpressurized AV, and simulate the AV function throughout a cardiac cycle. For geometric reconstruction purposes, dedicated software was created to acquire the 3-D coordinates of 21 anatomical landmarks of the AV apparatus in a systematic fashion. Measurements from ten 3D+t TEE datasets of normal AVs were assessed for inter- and intra-observer variability. These tests demonstrated mean measurement errors well within the acceptable range. Simulation of a complete cardiac cycle was successful for all ten valves and validated the novel schemes introduced to evaluate age-matched material properties and iteratively scale the unpressurized dimensions of the valves such that, given the determined material properties, the dimensions measured in vivo closely matched those simulated in late diastole. The leaflet coaptation area, describing the quality of the sealing of the valve, was measured directly from the medical images and was also obtained from the simulations; both approaches correlated well. The mechanical stress values obtained from the simulations may be interpreted in a comparative sense whereby higher values are indicative of higher risk of tearing and/or development of calcification.