Autonomous Neuro-Navigation System for Neurosurgical Robotics

Cranial neuro-navigation has become an integral part of contemporary neurosurgery, and is often practiced as an intensive manual process with the available clinical information. The neurosurgeons decide the craniotomy path referring to a non real time radiographic images, often leading to impromptu deviation in the surgical plan. The real time visualization of underlying structures are unavailable and the procedure highly rely on the expertise of the surgeon. Hence an alternate and effective method with minimal human intervention in the surgical space is needed for practicing neurosurgical craniotomy procedures. In this paper, an autonomous image guided neuro-navigation system is proposed for robotic neurosurgical procedures towards making the overall process efficient and error free. The presurgical and the intraoperative three dimensional data of the patient under investigation are acquired and a formal craniotomy path is decided by the neurosurgeon post co-registration of two modality of patient’s data. A novel registration work flow is incorporated in the autonomous system that require no reference implantable fiducials and is implemented in two stages. In the proposed registration scheme, the initial coarse alignment of 3D data is accomplished using local shape feature descriptors which is further refined by point to point registration. The developed navigation guidance system was then validated in an experimental setup with a needle fixed to a movable drilling head. The experimental results demonstrated successful positioning of the needle over the specified annotated point, with an acceptable accuracy, as referred and driven by 3D reconstructed model derived from the set of radiographic images.