Introduction: Cerebral vessel injury during thrombectomy may lead to complications. Measuring stresses and tracking displacements of model regions could inform safer thrombectomy practices. Methods: Silicone middle cerebral artery (MCA) models were imaged with nine off-the-shelf cameras (20 fps, 2160p). Multiple camera feeds were fed into 4D Gaussian Splatting to reconstruct dynamic 3D point-cloud representations of the vessel as thrombectomy is performed, which were converted into meshes for stress analysis using a linear-elastic spring network model. Fluorescent microspheres were placed on the vessel surface to improve contrast and tracking fidelity. The analysis was performed offline, with temporal resolution limited only by the cameras’ frame rate. Results: We analyzed a silicone MCA model under two conditions (one trial each) each using a standardized, device-driven pullback. Under Condition 1 (Toro catheter; Cervical ICA placement; M2-M1 stent), peak displacements relative to starting position were 2.5 mm at the bifurcation and 1.8 mm at the proximal M2 segment. In Condition 2 (Toro catheter; ICA terminus placement; M2-M1 stent), peak displacements were 0.6 mm and 0.3 mm, respectively. Preliminary stress analysis showed high-stress regions at the vessel neck connecting the loop segment to the branching vessel. Validation against geometric fiducials and a comparative stress analysis across thrombectomy techniques and different clot locations is ongoing and will be presented. Conclusion: A low-cost multi-camera pipeline can generate interpretable displacement measurements at clinically relevant landmarks and localize relative vessel-wall stress during thrombectomy in silicone vessel models. With material calibration and expanded sampling, this methodology could support validation studies and future comparisons across thrombectomy maneuvers.
Carmona et al. (Thu,) studied this question.