Brain shift in neuronavigation One of the major technical problems facing conventional neuronavigation techniques is brain shift. Because conventional navigation uses the patient’s preoperative imaging data. As the brain tissue is not a rigid body, brain deformation [29-30] (brain deformation), also known as drift, often occurs during actual surgery due to tissue biomechanical properties, gravity, intracranial pressure changes, cerebrospinal fluid loss, surgical operations, and anesthesia status. A summary of 1000 neurosurgery cases in the Department of Neurosurgery at Huashan Hospital showed [31]: dural shift of 2.80 ± 2.48 mm, cortical shift of 5.14 ± 4.05 mm, and tumor shift of 3.53 ± 3.67 mm, with the cerebral hemisphere surgery being the most dramatic. Brain shift errors lead to a decrease in the accuracy of neuronavigation localization using preoperative images, which interferes with the accuracy and safety of surgery and leads to postoperative tumor residuals or damage to normal neurovascular structures. Therefore, the study of new techniques to correct brain shift errors has become a current hot topic in the field of neuronavigation surgery. In general, brain shift can be solved in three ways: (1) microcatheter localization technique; (2) computational model-updated image technique; (3) intraoperative real-time imaging technique.
1. Microcatheter localization technique Before the dura mater is cut, a micro-silicone tube (1-2 mm in diameter) is placed around the lesion under the guidance of neural navigation. After the dura is cut, although the cerebrospinal fluid is lost or the brain is displaced during the resection of the lesion, the microcatheter also moves with it, and the surgeon can further complete the surgical operation under the guidance of the microcatheter. Huashan Neurosurgery created this method in 1999, and it has been confirmed by long-term clinical practice that it is simple, economical and effective, but the shortcoming is that this technique is rougher in positioning.
2.Model correction technique The brain shift is compensated and corrected by correction software technique. At present, there are mainly three kinds of models: mathematical model (such as B-sample model), physical model (such as linear elastic model and solidification theory model) and brain deformation atlas (BDA) method. The core is a non-rigid registration method based on a computational model [32]. In the previous study, our group designed a linear elastic physical model and a mathematical model based on the thin-slab-like algorithm to simulate intraoperative brain tissue deformation more accurately, which is a simple, fast and reliable way to correct brain deformation errors.
(1) Thin-slab-like mathematical model Based on the research of some foreign scholars [33], this group predicts the brain functional image deformation by improving the 3D image nonrigid alignment algorithm of thin-slab-like strips for the solution of internal deformation (Invention patent application number: 200910047537.2). The mathematical model of thin slab-like strips is applied to interpolate the deformation at any location inside the brain tissue by corresponding to the change in the position of anatomical marker points. The predicted deformed functional brain images (BOLD and DTI) are then fused with intraoperative MRI structural images to resolve the functional brain localization errors caused by brain displacement [34]. In this study, we used preoperative MRI images and intraoperative MRI images as preoperative and intraoperative data fields, respectively, and the deformation of anatomical marker points was obtained by the alignment of preoperative and intraoperative data fields, thus avoiding the errors caused by physical models, and the results of both animal experiments and clinical trials verified the good accuracy of predicted deformation. In this study, real-time functional neurological navigation based on low-field intensity intraoperative MRI has been achieved, which is a preliminary step to overcome this international problem.
(2) Linear-elastic physical models Physical models can constrain the movement of brain tissue through its biomechanical properties (e.g., tissue elasticity, water pressure transfer value, etc.), and are therefore also called biomechanical models. Compared with mathematical models, the advantages of these models are that they can significantly reduce computational effort, do not require large samples, have reliable accuracy, and can be easily applied in clinical settings. Our group has developed a linear elastic physical model (Patent No. ZL200410024847.X 08/23/2006) to simulate intraoperative brain tissue deformation more accurately.
Animal experiments confirmed [35, 36] that the mean prediction error of this linear elastic physical model is <1 mm (0.97±0.44 mm); the correction accuracy is as low as 56.5% and as high as 90.0%, with a mean of 68.0±9.6%. With this model, we wrote the brain shift correction software, 3D Imageâ which can be loaded on the platform of FDM Excelim-04? a domestic neuronavigation system (developed by the Digital Medical Center of Fudan University). We confirmed through clinical trials that the model predicts reliable results and can significantly improve the accuracy and safety of neuronavigation surgery.
(3) drive the model based on the relationship between the surface of the operative field and the deep deformation, and finally generate high-resolution and predictive accuracy brain shift correction images.
3.Intraoperative imaging techniques Intraoperative imaging techniques are currently more mature techniques, including CT, ultrasound and MRI imaging techniques. The earliest techniques used for intraoperative imaging were CT and ultrasound, which were first reported by Shalit (1979) and Rubin (1980), respectively. Although CT has been improved recently and has good resolution, especially for bone, it is still not as good as MRI for soft tissues, and because CT is radioactive, it can be harmful to the human body when working in this environment for a long time. Intraoperative ultrasound technology has recently developed rapidly and can be used for 2D and 3D imaging, but its resolution is still inferior to that of CT or MRI, and the penetration ability of ultrasound is inversely proportional to the resolution, i.e., the penetration decreases when the resolution increases. Therefore, due to these shortcomings, the application of intraoperative CT and intraoperative ultrasound is limited and not promoted. Therefore, intraoperative magnetic resonance imaging (iMRI) is now more commonly used to correct brain displacement.