ABSTRACT Image‐guided neurosurgery system (IGNS) has become an indispensable component of modern precise neurosurgery. Image‐to‐patient registration plays a key role in IGNS, as it directly impacts accuracy and security of the surgery. We present a novel surface‐based image‐to‐patient registration method for IGNS, which is feature‐free and performs direct registration. First, the bounding boxes of the surfaces are computed, and a convex plane‐based approach is developed to obtain an initial posture close to the desired value. Then, a constraint‐based dimensionality reduction strategy is proposed to identify the most optimal posture for coarse registration. Finally, the iterative closest point (ICP) algorithm is employed to refine and generate the final transformation. This registration method is evaluated using the phantom models. The results demonstrate that our method produces transformations comparable to the currently widely used method. The proposed feature‐free method exhibits the potential to facilitate IGNS with sufficient precision.
Li et al. (2026) studied this question.