A feasibility study of several 3D printing methods for applications in epilepsy surgery

To describe the process of three-dimensional printing in epilepsy surgery using three different methods: low-force stereolithography (SLA), filament deposition modeling (FDM), and Polyjet Stratasys, while comparing them in terms of printing efficiency, cost, and clinical utility. MRI and CT images of patient anatomy have been limited to review in the two-dimensional plane, which provides only partial representation of intricate intracranial structures. There has been growing interest in 3D printing of physical models of this complex anatomy to be used as an educational tool and for surgical visualization. One specific application is in epilepsy surgery where there are challenges in visualizing complex intracranial anatomy in relation to implanted surgical tools.

MRI and CT data from patients with refractory epilepsy from a single center that underwent surgery are converted into 3D volumes, or stereolithography files. These were then printed using three popular 3D printing methods: SLA, FDM, and Polyjet. Faculty were surveyed on the impact of 3D modeling on the surgical planning process.

All three methods generated physical models with an increasing degree of resolution, transparency, and clinical utility directly related to cost of production and accurate representation of anatomy. Polyjet models were the most transparent and clearly represented intricate implanted electrodes but had the highest associated cost. FDM produced relatively inexpensive models that, however, were nearly completely opaque, limiting clinical utility. SLA produced economical and highly transparent models but was limited by single material capacity.

Three-dimensional printing of patient-specific anatomy is feasible with a variety of printing methods. The clinical utility of lower-cost methods is limited by model transparency and lack of multi-material overlay respectively. Polyjet successfully generated transparent models with high resolution of internal structures but is cost-prohibitive. Further research needs to be done to explore cost-saving methods of modeling.