1243 Realistic Visualization for Neurosurgical Simulation and Education

INTRODUCTION: Artificial intelligence (AI)-based segmentation technology has enhanced the accessibility and ease of surgical simulation. Recently, realistic visualization via three-dimensional computer graphics (3DCG) has gained importance for better surgical simulation and education. Visualizing microanatomy, including the cerebellar sulci, remains challenging with current technology. Voxel intensity-based material attaching (VIMA) on surface rendering, which adds the voxel intensity of the original medical image to the 3DCG as a vertex color, may improve the visualization of the cerebellar sulci and enhance the reality of the cerebellar 3DCG. METHODS: To create 3DCG, 10 cases of contrast-enhanced fast imaging employing steady-state acquisition were processed using the following three different techniques: AI-based segmentation with VIMA (AI with VIMA), AI-based segmentation without VIMA (AI only), and manual segmentation without VIMA (non-AI). Three experts performed cerebellar sulci counts on each 3DCG. The inter-rater reliability of cerebellar sulci counts was evaluated using the intraclass correlation coefficient (ICC), and then cerebellar sulci counts were compared among the three techniques. 3DCG creation time between AI and non-AI methods was also compared. RESULTS: For all techniques, the ICC was >0.9, and the inter-rater reliability of the cerebellar sulci counts was excellent. Significant differences were observed between AI with VIMA and AI only (p<0.001) as well as the non-AI method (p=0.042). No significant difference was found between AI only and non-AI (p=0.22). AI also had a significantly shorter 3DCG creation time (p<0.001). CONCLUSIONS: A new 3D visualization technique was developed. The 3DCG created from AI with VIMA visualized more cerebellar sulci without the time-intensive work, contributing to improved realistic visualization.