Registration Using Cortical Surface for Microscope-based Image-guided Neurosurgery

Microscope-based Image-guided Neurosurgery seeks to establish an accurate correlation between preoperative images and the current surgical scene under the microscope through rigid registration. Information, drawn from diagnostic image data, including computerized tomography (CT) and magnetic resonance images (MRIs), can be projected into the operating microscope, overlaying the surgical field in the correct position, scale, and orientation. In an effort to improve the efficiency and reliability of Microscope-based Image-guided Neurosurgery, three major improvements of the system are proposed in this dissertation proposal. First, a real time correction scheme that employs motionless markers on the bony rim of the craniotomy site is used to combat the system perturbation. The accuracy of camera calibration parameters is actively maintained by reinforcing the rigid registration between their 3-D locations in the world and their 2-D images under microscope. Second, cortical vessels are proposed to replace implanted markers on the patient’s scalp as fiducials to register the preoperative images to the operating space. The objective is to automate the step of patient registration, reduce the invasiveness, and achieve accuracy. Third, the movement of the cortical surface will be tracked during the operation, so that the surface displacement data can be used to guide the finite element model to recover the brain shift. Stereopsis will be used to digitize the cortical surface and vessels in the world. In searching for an accurate and robust solution for the correspondence problem in stereopsis, a range of matching constraints will be implemented and quantitatively tested.