NI-18 * MULTIMODAL NAVIGATION IN ENDOSCOPIC TRANS-SPHENOIDAL RESECTION OF PITUITARY TUMORS USING IMAGE-BASED VASCULAR AND CRANIAL NERVE SEGMENTATION: A PROSPECTIVE VALIDATION STUDY

Abstract

Trans-Sphenoidal surgery (TSS) is a well-known approach for treatment of pituitary tumors. However, in inexperienced hands, the risk of lateral misdirection and vascular damage, intraoperative CSF leakage, and optic nerve injury are all well-known complications of this procedure. This prospective study was conducted to validate the accuracy of image-based segmentation in localization of neurovascular structures during TSS. METHODS: Eight patients with pituitary tumor underwent preoperative 3TMRI, which included thin sectioned 3D space T2, 3D Time of Flight and MPRAGE sequences. Images were reviewed by an expert independent neuroradiologist. Imaging sequences were loaded in BrainLab iPlanNet (6/8) and Stryker (2/8) for segmentation and pre-op planning. After patient registration to the intra-op neuronavigation system and surgical exposure, each segmented neural or vascular element was validated by manual placement of the navigation probe. The pulses of the bilateral ICA were confirmed using micro-Doppler. RESULTS: Pre-operative segmentation of the ICA and cavernous sinus matched with the intra-operative endoscopic and micro-Doppler findings in all cases (Dice-coefficient =1). This information reassured surgeons regarding the lateral extent of bone removal at the sellar floor and the limits of lateral explorations. Perfect correspondence between image-based segmentation and endoscopic view was also found at the surface of the tumor and tumor-normal gland interfaces. This helped in preventing unnecessary removal of the normal pituitary gland. Image-guidance helped surgeon to localize the optic nerve and chiasm in 63% of case and Diaphragma sella in 50% of cases, which helped to determine the limits of upward exploration and decrease the risk of CSF leakage. CONCLUSION: Image-based pre-operative vascular and neural element segmentation especially with 3D reconstruction is highly informative preoperatively and helps young and inexperienced neurosurgeons to prevent vascular and neural injury during trans-Sphenoidal surgeries. This novel preliminary study is encouraging for future prospective intraoperative validation with larger numbers of patients.