AI-Powered Segmentation Revolutionizes Scaffold Design in Regenerative Dentistry

The era of 3D printing of biocompatible personalized scaffolds has arrived, and Cone Beam Computed Tomography (CBCT) is essential for 3D reproduction of individualized human anatomy. When designing the shape of the personalized scaffold, the starting point is typically the CBCT scan, which must first be segmented to define the complementary shape of the scaffold. In the past, this was usually a lengthy manual segmentation process that could take hours. Today, artificial intelligence-based software can perform automatic segmentation of the various structures in the maxillo-facial region directly from CBCT data in seconds. This study presents a novel workflow comparing manual segmentation in Invivo (Anatomage, San Jose, CA, USA) with AI-automated segmentation in Diagnocat (Miami, FL, USA). In 24 cases, the time required for segmentation were compared and evaluated with a paired t-test. This revealed a statistically significant difference in segmentation time between the two groups, with the AI-driven analysis being significantly faster. The difference in average segmentation time between manual (36.03 minutes) and AI-driven analysis (4.96 minutes) showed that AI-driven analysis was on average more than five time faster than manual segmentation. AI-driven analysis reduces segmentation time by 86.17% compared to manual segmentation of CBCT. This means that AI-driven analysis can save clinicians a lot of time. The presented feasibility of AI-automated workflow with STL (Standard Triangle Language) output models suitable for 3D modeling of scaffold shapes - complementary to individual anatomy in Meshmixer (Autodesk, San Rafael, CA, USA), were suitable for 3D printing with hydroxyapatite. This has significance for various workflows in regenerative dentistry.