Nonlinear MR-US Registration for Image Guided Neurosurgery of Brain Tumours
The broad aim of this project is to develop and evaluate nonlinear registration techniques that account for discontinuities in data from tissue resection in order to accurately reflect the deformed brain. These registration procedures will improve patient-image alignment during surgery and enable correction of preoperative data to reflect the intraoperative physical reality of brain tissues. This will result in surgical interventions performed with greater accuracy and confidence, reducing the risk of complications and enabling more complex procedures within the framework of IGNS. The alignment of iUS to magnetic resonance (MR) images is a challenging task due to the widely different nature and quality of the two modalities. While voxel intensity of both modalities is directly dependent on tissue type, US has an additional dependence on probe orientation and depth that can lead to intensity non-uniformity due to the presence of acoustic impedance transitions. Preoperative MR images allow for accurate and precise identification of tissue types, anatomical structures and a variety of pathologies such as cancerous tumours. Intraoperative US images are generally limited to displaying lesion tissue with an associated uncertainty regarding its boundary, along with a few coarsely depicted structure boundaries. MR – US registration techniques to correct for brain shift have recently been developed, based on gradient orientation alignment, in order to reduce the effect of the non-homogeneous intensity response found in iUS images. While accurate, this technique is optimal for small shifts and uses a rigid registration followed by a linear elastic deformation model; a simplified model of the brain’s mechanical properties. Current registration techniques take 10 – 20 minutes of computational time during surgery (due to iUS reconstruction from 2D slices), making them less useful because they can delay surgery. Current work in Dr. Collins’ laboratory has developed a linear group-wise slice-to-volume registration procedure in the context of image guided spine surgery4. This slice-to-volume technique has many advantages most important of which is a decrease in computational time. Expanding these techniques to IGNS of the brain will help eliminate some of the time constraints currently associated with using registration procedures during surgery.
Ian Gerard , McGill University
Partners and Donors