Shape-Based Dynamic Contact Force Estimation for Guidewires in Endovascular Interventions

Force perception is essential for safe robotic endovascular navigation, but distal force sensors are difficult to integrate into submillimeter guidewires. This letter presents a topology-constrained, shape-based framework for estimating guidewire–vessel contact forces during guidewire navigation. Reconstructed guidewire shapes provide the contact number, locations, and spans as geometric priors, reducing the inverse problem to force amplitude estimation under a Cosserat-rod model. A two-phase module identifies an effective friction coefficient from proximal force measurements and compensates friction-induced axial force transmission for dynamic contact force estimation and distal tip force inference. The framework was evaluated in Abaqus FEA and controlled phantom experiments. FEA produced local contact force RMS errors of2.14\,mN(6.73%) under frictionless conditions and3.22\,mN(6.47%) under frictional conditions. In vitro experiments showed shape reconstruction RMS errors≤ 0.24\,mmin all phantoms. 2-D static point-force tests gave RMS errors of0.071\,N(8.30%) for the constant-stiffness Nitinol wire and0.031\,N(8.64%) for the variable-stiffness shaft, while dynamic proximal reaction force errors ranged from0.73\,mNto3.56\,mNacross six guidewire–phantom cases. These results indicate the feasibility of sensor-free, friction-aware guidewire force estimation across different validation settings.