Automatic non-rigid registration of preoperative MRI and intraoperative US for US-guided neurosurgery - A preliminary study

Image guidance aids neurosurgeons in making critical clinical decisions to achieve a safe maximal resection of diseased tissue. The brain however undergoes significant nonlinear structural deformation due to multiple factors, including dura opening, fluid shifts, and tissue resection. While ultrasound (US) has shown promise in monitoring brain deformation and as a navigation image source in its own right, magnetic resonance imaging (MRI) generally provides better contrast between tumor and brain tissue and pre-operative imaging can provide functional and connectivity information not available in US. Mapping pre-operative MRI (pMRI) onto intraoperative US (iUS) is therefore desirable for multimodal neurosurgical guidance. In this work, we present a novel approach that uses T1 intensity gradients to non-rigidly register pMRI to iUS. The method was validated on 22 datasets from the MICCAI CuRIOUS Challenge 2018 and an average landmark error of 4.89mm was achieved.