Tracking Cortical Surface Deformation Using Stereovision

Abstract

Tracking brain deformation is important for the understanding of brain biomechanical properties. However, accurate deformation tracking may be challenging especially in vivo. We present a completely noninvasive technique to track cortical surface deformation using intraoperative stereovision during open cranial neurosurgery. A sequence of stereo images was acquired to capture motion of the exposed cortical surface due to blood pressure pulsation during multiple respiration cycles for patients undergoing brain tumor resection surgery. Rigid registration was performed between the first and subsequent frames using features outside the craniotomy to compensate for accidental motion of the surgical microscope. A nonrigid registration based on optical flow was performed next between the first and subsequent frames. A reference image was then generated using pixel displacements averaged across an integer multiple of respiration cycles to serve as a reference state based on which a dense displacement field was determined for each image frame. The resulting displacement field was then locally smoothed to minimize noise, and was further spatially differentiated to compute in-plane surface strain using deformation gradient. The technique offers an effective approach to track deformation of soft tissue surface as long as sufficient tracking features are available, which is useful for soft tissue biomechanical characterization.