Robot-Assisted Transcranial Magnetic Stimulation Guided by Cortical Topography

Transcranial magnetic stimulation (TMS) is widely used for functional brain mapping and neuromodulation, but conventional neuronavigation systems typically rely on simplified volumetric brain models that do not fully capture subject-specific cortical geometry. We developed and evaluated a novel cortical-topography-guided framework for improving TMS planning, coil alignment, and stimulation reliability. The framework integrates an enhanced head-MRI co-registration strategy based on the iterative closest point (ICP) algorithm using a three-dimensional cortical mesh reconstructed from structural MRI. This approach was designed to improve neuronavigation by more accurately aligning the participant's head with the MRI-derived cortical anatomy, thereby reducing co-registration error. Robot-assisted TMS (RA-TMS) was used to preserve the planned coil pose relative to cortical targets. Registration accuracy, coil alignment error, and motor-evoked potential (MEP) responses at the hand area of the primary motor cortex (M1) were evaluated and compared between RA-TMS and conventional manual TMS approach, which did not consistently preserve cortical topography-guided stimulation. ICP-based co-registration reduced the registration error between the head and MRI model from 5.83 ± 1.04 mm to 1.07 ± 0.03 mm. Compared with manual alignment, RA-TMS reduced translational error from 14.79 ± 8.03 mm to 5.22 ± 1.68 mm and rotational error from 7.01° ± 2.29° to 1.09° ± 0.33° (p < 0.001). RA-TMS also produced larger and more consistent MEP responses than manual TMS (0.27 ± 0.05 mV vs. 0.03 ± 0.01 mV, p < 0.001). Incorporating cortical topography-guided planning, combined with refined co-registration and robotic stabilization, improves coil alignment accuracy and the reliability of stimulation outcomes.