Biomechanical study of probe-tissue interactions during deep brain stimulation: Force measurement and channel...

Understanding the mechanical interactions between surgical probes and brain tissue is essential for optimizing procedures such as deep brain stimulation (DBS). In this study, agar gel phantoms were used as brain tissue surrogates for their well-characterized mechanical properties and extensive use in neurosurgical modeling. Systematic experiments were conducted to quantify the insertion and withdrawal forces of DBS probes, and the evolution of probe-induced channels was analyzed using synchronized high-speed imaging and force measurements. Key parameters, such as peak insertion force, were extracted from filtered curves showing the relationship between force and depth as well as force and time. Classical physical models, such as the Hertz and Fung equations, characterized the force response in the linear regime and in regimes with weak nonlinearity, while a hybrid physics-guided residual neural network (PGNN) was applied to capture complex and highly nonlinear interactions. Our results show that the force response exhibited time-dependent behavior: insertion force showed clear velocity dependence, whereas withdrawal force was predominantly described by a velocity-independent friction term over the tested range. Probe speed and gel-recovery dynamics nevertheless influenced channel closure. Channel measurements revealed that the residual channel is consistently smaller than the probe diameter, which can be attributed to the combined effects of elastic recovery, viscous flow, and hydration. Both probe speed and depth were found to significantly influence the dynamics of channel closure. Model fitting demonstrated that classical models can adequately describe the force response in specific regimes, but the hybrid PGNN model improves prediction accuracy for complex mechanical interactions. Overall, this work offers new insights into phase-specific probe-material interaction mechanics in a controlled homogeneous surrogate, and the integrated experimental and modeling framework developed here provides a baseline dataset for future studies of DBS-relevant insertion mechanics and model refinement.