Near Real-time Computer Assisted Surgery for Brain Shift Correction using Biomechanical Models

摘要

Conventional image-guided neurosurgery relies on preoperative images to provide surgical navigational information and visualization. However, these images are no longer accurate once the skull has been opened and brain shift occurs. To account for changes in the shape of the brain caused by mechanical (e.g. gravity-induced deformations) and physiological effects (e.g. hyperosmotic drug-induced shrinking, or edema-induced swelling), updated images of the brain must be provided to the neuro-navigation system in a timely manner for practical use in the operating room. In this study, a novel preoperative and intraoperative computational processing pipeline for near real-time brain shift correction in the operating room was developed to automate and simplify the processing steps. Preoperatively, a computer model of the patient’s brain with a subsequent atlas of potential deformations due to surgery is generated from diagnostic image volumes. In the case of interim gross changes between diagnosis, and surgery when re-imaging is necessary, our preoperative pipeline can be generated within one day of surgery. Intraoperatively, sparse data measuring the cortical brain surface is collected using an optically-tracked portable laser range scanner. This data is then used to guide an inverse modeling framework whereby full volumetric brain deformations are reconstructed from pre-computed atlas solutions to rapidly match intraoperative cortical surface shift measurements. Once complete, the volumetric displacement field is used to update, i.e. deform, preoperative brain images to their intraoperative shifted state. In this work, 5 surgical cases were analyzed with respect to the computational pipeline and workflow timing. With respect to post-cortical surface data acquisition, the approximate execution time was 4.5 minutes. The total update process which included positioning the scanner, data acquisition, inverse model processing, and image deforming was approximately 11 - 13 minutes. In addition, easily implemented hardware, software, and workflow processes were identified for improved performance in the near future.