Intraoperative Near-Infrared Fluorescence Imaging With Second-Window Indocyanine-Green Offers Accurate and Real-Time Correction for Brain Shift and Anatomic Inconsistencies in Framless Neuronavigation

Abstract

Abstract INTRODUCTION Neuronavigation allows neurosurgeons to localize intracranial structures in 3D space and has been shown to have positive impacts on patient survival in neuro-oncology surgery. However, its major limitation is the intraoperative brain-shift phenomenon, in which the brain moves during surgery due to physical, surgical, and/or biological factors, invalidating the preoperative registration and leading to inaccuracies in the cranium. One reliable way to account for brain-shift intraoperatively may be to use tumor-targeting near-infrared fluorophores, such as ICG accumulating in neoplastic tissue (ie, Second-Window ICG), which offers real-time visualization of tumors through normal cortex and dura. METHODS Patients undergoing craniotomy for primary resection of intracranial tumors were enrolled under an ongoing clinical trial investigating the efficacy of SWIG. For this analysis, retrospective data were collected on patients in whom neuronavigation was used to plan the craniotomy in such a way as to place the tumor in the center of the craniotomy (ie, parasagittal tumors, skull-base tumors, and large/asymmetrical tumors were excluded). During surgery, near-infrared imaging was performed before and after durotomy to localize the gross tumor. Image analysis was performed to measure the deviation between the craniotomy center and the center of the tumor as seen with near-infrared fluorescence. RESULTS A total of 63 patients (24 high-grade-gliomas, 12 meningiomas, 25 metastases, 2 others) were included in this preliminary analysis. Neuronavigation demonstrated a median deviation of 22.4% (range 7.7%-44.4%) relative to craniotomy size. Patient position was a significant predictor of neuronavigation inaccuracy, with the prone position having significantly higher inaccuracy (28.5 ± 8.8%) compared to the supine (19.3 ± 9.2%, P-value = .015) or the lateral (17.9 ± 6.6%, P-value = .012) positions. Additionally, the neuronavigation device used and postgraduate training level of the residents performing the registration trended towards significance on multivariate analysis. In contrast, near-infrared fluorescence perfectly delineated the tumor in all cases. CONCLUSION We demonstrate that near-infrared fluorescence imaging offers more accurate localization of intracranial tumors compared to frameless neuronavigation. Near-infrared imaging could potentially be used to adjust neuronavigation registrations intraoperatively to enhance accuracy. Further prospective studies with distance measurements could better explore this potential benefit of intraoperative near-infrared imaging.