Advanced cranial navigation

Abstract

Neurosurgery is performed with extremely low margins of error. Surgical inaccuracy may have disastrous consequences. The overall aim of this thesis was to improve accuracy in cranial neurosurgical procedures by the application of new technical aids. Two technical methods were evaluated: augmented reality (AR) for surgical navigation (Papers I-II) and the optical technique of diffuse reflectance spectroscopy (DRS) for real-time tissue identification (Papers III-V).

Minimally invasive skull-base endoscopy has several potential benefits compared to traditional craniotomy, but approaching the skull base through this route implies that at-risk organs and surgical targets are covered by bone and out of the surgeon’s direct line of sight. In Paper I, a new application for AR-navigated endoscopic skull-base surgery, based on an augmented-reality surgical navigation (ARSN) system, was developed. The accuracy of the system, defined by mean target registration error (TRE), was evaluated and found to be 0.55±0.24 mm, the lowest value reported error in the literature.

As a first step toward the development of a cranial application for AR navigation, in Paper II this ARSN system was used to enable insertions of biopsy needles and external ventricular drainages (EVDs). The technical accuracy (i.e., deviation from the target or intended path) and efficacy (i.e., insertion time) were assessed on a 3D-printed realistic, anthropomorphic skull and brain phantom; Thirty cranial biopsies and 10 EVD insertions were performed. Accuracy for biopsy was 0.8±0.43 mm with a median insertion time of 149 (87-233) seconds, and for EVD accuracy was 2.9±0.8 mm at the tip with a median angular deviation of 0.7±0.5° and a median insertion time of 188 (135-400) seconds.

Glial tumors grow diffusely in the brain, and patient survival is correlated with the extent of tumor removal. Tumor borders are often invisible. Resection beyond borders as defined by conventional methods may further improve a patient’s prognosis. In Paper III, DRS was evaluated for discrimination between glioma and normal brain tissue ex vivo. DRS spectra and histology were acquired from 22 tumor samples and 9 brain tissue samples retrieved from 30 patients. Sensitivity and specificity for the detection of low-grade gliomas were 82.0% and 82.7%, respectively, with an AUC of 0.91.

Acute ischemic stroke caused by large vessel occlusion is treated with endovascular thrombectomy, but treatment failure can occur when clot composition and thrombectomy technique are mismatched. Intra-procedural knowledge of clot composition could guide the choice of treatment modality. In Paper IV, DRS, in vivo, was evaluated for intravascular clot characterization. Three types of clot analogs, red blood cell (RBC)-rich, fibrin-rich and mixed clots, were injected into the external carotids of a domestic pig. An intravascular DRS probe was used for in-situ measurements of clots, blood, and vessel walls, and the spectral data were analyzed. DRS could differentiate clot types, vessel walls, and blood in vivo (p<0,001). The sensitivity and specificity for detection were 73.8% and 98.8% for RBC clots, 100% and 100% for mixed clots, and 80.6% and 97.8% for fibrin clots, respectively.

Paper V evaluated DRS for characterization of human clot composition ex vivo: 45 clot units were retrieved from 29 stroke patients and examined with DRS and histopathological evaluation. DRS parameters correlated with clot RBC fraction (R=81, p<0.001) and could be used for the classification of clot type with sensitivity and specificity rates for the detection of RBC-rich clots of 0.722 and 0.846, respectively. Applied in an intravascular probe, DRS may provide intra-procedural information on clot composition to improve endovascular thrombectomy efficiency.