A Comparative Study of Industry and Open-Source Efforts for 3D Visualization, Pre- and Intraoperative Planning, and 3D Printing of Skull Base Tumors: A Case Report

We report exhaustive 3D modeling efforts for the purposes of pre-operative planning and intraoperative navigation in the case of a skull-base epidermoid tumor resection. Our goals were to understand the relative utility of each approach to modeling for skull-base tumors while considering the effort and technical knowledge required to produce valuable results. Preoperatively, thin-cut, “simulation protocol” MR (T1 pre-/postcontrast, CISS, MR-A) and CT/CT-A scans were obtained to maximize contrast between modeled structures, which included tumor, bone, relevant vasculature, and cranial nerve bundles (including the 5th, 7th and 8th, and 9th, 10th, and 11th nerves). Maximizing structural contrast in thin-cut, “simulation protocol” scan sequences improves and in some cases makes possible downstream segmentation of the images into 3D models of each structure. These scans were used as input to a variety of modeling platforms, including (1) Brainlab’s iPlan software linked to their intraoperative navigation platform, from which bone and vasculature were auto-segmented, and tumor was painted using the included “smart brush” feature; and (2) Surgical Theater’s Surgical Navigation Advanced Platform (SNAP), in which transfer functions were optimized to volume render bone, vasculature, and tumor, followed by manual painting of cranial nerves. Both platforms were made available during surgery using navigation linked to Brainlab’s standard navigation platform. A third, more exhaustive modeling effort was achieved through the use of a suite of available open-source tools, including 3DSlicer and associated modules, Seg3D2, and in-house segmentation tools based on SimpleITK and ITK (made available open-source on our Web site). The technique used for each individual structural segmentation varied depending on location, complexity, and software, but included interactive methods all the way down to fully-manual expert-driven segmentation (e.g., “painting,” esp. in the case of cranial nerves). Unlike most industry-available platforms, the exhaustive structural modeling of bone, tumor, cranial nerves, and vasculature achieved here results in the generation of open-format shape files (e.g., VTK or STL) appropriate for analysis in any platform accepting these formats (of which there are thousands). We specifically report our experience with a novel interactive visualizer developed in-house based on VTK, and multi-component multi-color 3D printing on an industrial-scale 3D Systems Projet 660 printer. We report advantages and disadvantages of each approach described here, pitfalls encountered, and consider the associated efforts and costs, to develop a prototype best-practices guide to 3D modeling for tumors of the skull base.