Inverse approaches for the mechanics of 3D-printed surgical phantoms : Identification and kinematic analysis from intraoperative imaging via projection-based digital volume correlation

Recent additive manufacturing techniques have enabled the fabrication of tissue-mimicking materials with widespread applications in the medical field, particularly for clinical simulation. In the context of EndoVascular Aneurysm Repair (EVAR), patient-specific phantoms of abdominal aortic aneurysms are produced to enhance surgical training. A large proportion of these models are manufactured using the PolyJet process, which enables the printing of property gradient structures by combining soft and rigid photopolymer materials. Although aortic 3D-printed models achieve strong anatomical fidelity, their mechanical reliability remains limited. Yet, reproducing realistic mechanical behaviour in the physical models is a prerequisite for their implementation in clinical settings. A first step toward improving phantom reliability is to characterize the stress and strain fields experienced during EVAR training.EVAR procedures, as well as their training counterparts, are performed in hybrid operating rooms equipped with cone-beam computed tomography, enabling both three-dimensional (3D) pre- and post-operative imaging and live two-dimensional (2D) fluoroscopy. This multimodal acquisition framework constitutes a valuable source of data for capturing the aortic model kinematics. While standard Digital Volume Correlation (DVC) can extract kinematic information from volumes, it does not benefit from the additional 2D fluoroscopy data. This thesis investigates the recent projection-based digital volume correlation approach to evaluate 3D kinematics by combining volumetric images with a series of 2D radiographs acquired at varying angles and loading conditions throughout surgery, thereby yielding richer temporal information. This framework has the potential to be extended to in operands endovascular navigation in real patients undergoing EVAR, where resolving aortic motion is critical for accurate image fusion.To assess the stress state of the surgical phantom, a reliable constitutive model accounting for surgery-specific loading conditions is required. This is also addressed by the development of multiaxial mechanical tests under controlled conditions to identify hyperelastic and time-dependent constitutive laws for PolyJet materials. The identification strategy relies on full-field measurements combined with inverse methods, namely integrated digital image correlation and finite element model updating. Numerical approaches for sensitivity optimisation are proposed for each identification problem.By integrating complementary experimental and numerical frameworks, this thesis provides new insights into improving the mechanical reliability of 3D-printed anatomical phantoms, advancing their use for surgical training, complex case rehearsal, and perioperative procedure refinement.