A Study On Correcting Brain Retraction In Image-Guided Neurosurgery

Image Guided Neurosurgery Systems (IGNS) are increasingly being used in the Operating Room (OR) and having showed improving qualities of surgical visualization and navigation as well as reducing the postoperative tumor residual.Brain retraction may cause large inconsistencies between images and real anatomy, which is one of the major source of the error in IGNS which was initially built based on preoperative images. Therefore, brain retraction correction is an important intraoperative clinical application. Brain retraction involves topological discontinuity, which has been proven to be more challenging to correct, especially to big brain retraction deformaion. Thus, related studies have been limited. Most studies employed FEM-based biomechanical models to warp preoperative images to reflect brain retraction. However, it is difficult for FEM to deal with tissue discontinuity. IMR scanning in most studies is still required to calculate BCs, so it may unnecessarily prolong operation time and create extra expenses for patients, which makes it unlikely to become widely available in clinical applications. Moreover, most studies only gave a qualitative evaluation on the method of brain retraction correction. The whole thesis can be divided into the following five parts:1. A framework was brought forward for brain retraction correction in which the biomechanical model was used and deformed. Based on the observation of the biomechanical response of the brain tissue, the linear elastic model and hyper-viscoelastic model were chosen to model brain retraction.2. Our linear biomechanical model was based on eXtended Finite Element Method (XFEM), which adds extra degrees of freedom to nodes related to discontinuity. This improvement compared with FEM makes mesh adaptation or remeshing unnecessary.3. In the work presented here, the application of LRS in neurosurgery can track sufficient surface points of the retractors especially for the inner sub-surface of retractors inserted into the brain and measure intraoperative brain retraction surfaces in an automatic and rapid fashion instead of using iMR images.4. We performed a brain phantom experiment involving the retraction of tissue to test the framework brought forward above. The displacements of beads embedded in the brain phantom between measured and predicted bead displacement were compared to evaluate numerical performance. The forecast error varies between 0.0 and 2.0 mm (mean 1.19 mm) and the correction accuracy between the forecasted and actual results varies between 52.8% and 100%(mean 81.4%). The verification experiments have been implemented on one dead swine brain and seven 3-month old live swines. The purpose of this application was to verify the effectiveness of this framework which was much close to the scenario of the clinical use. Based on the observation of the biomechanical response of the brain tissue, we replaced the linear elastic model by hyper-viscoelastic model, and then put the updated framework into use on the live swines. The average forecast error of seven swine was 0.42±0.16 mm and the average correction accuracy was 77.65±5.68%. The results of experiments showed that the retraction deformation predicted by the present framework agrees well with the observation in OR intraoperatively.5. We integrated our brain-retraction-correction framework into the excelim-04, a type of IGNS, which currently can correct the brain shift after opening the dura mater. This update could make the excelim-04 have the capbility of correcting both brain deformation after opening the dura mater and brain retraction intraoperatively. Here, we describe what hardware will be needed and the necessary workflow of the excelim-04. The results demonstrate that complex surgical events (tissue retraction) can be incorporated intraoperatively into the model-updating process for brain retraction compensation.In conclusion, the results of verification experiments demonstrate that the presented framework has capability to recover brain retraction and improves the navigation accuracy of IGNS. The brain-retraction-correction framework can be integrated into IGNS. With the research in this thesis as the major contents, one paper was published in English and one paper in English has been submitted to one SCI indexed journal too.