Additive manufacturing of patient-specific implants using an AI-based digital workflow in oromaxillofacial surgery

Abstract text (incl. figure legends and references)

A possible loss of function of the eye in the context of a fracture of the orbital floor, which is one of the most common bony injuries after facial trauma, requires an optimized patient-specific therapy. The treatment with PSI in manual-additive manufacturing technology offers a high quality of care, but is associated with a high personnel, time and thus also financial expenditure. The aim of the DigiMed project is to develop a digital-additive manufacturing process. Starting with AI-based geometry generation via individualized manufacturing, this is expected to bring about a 50% reduction in the development and manufacturing time of PSIs and a 40% reduction in manufacturing costs. In the first step, AI-based pre-processing of medical image data of the orbital floor defect is performed. Automatic segmentation is performed by a specially designed Convolutional Neural Network (CNN). An approach based on a Statistical Shape Model (SSM) is presented for the virtual reconstruction of the defective orbital anatomy. After the segmentation is done, another algorithm is used for the automatic generation of the implant design. The patient-specific implant design requirements are defined by landmarks. The CNN can automatically recognize the landmarks and combine them with medical, manufacturing and design requirements. Subsequently, process strategies are developed for the additively manufactured implants that allow adaptive adjustment of the process parameters to the implant design. These should lead to an optimization of the quality, of the implants produced by AM manufacturing. For the validation of density and biocompatibility, test specimens were fabricated and corresponding biocompatibility tests were performed. The sample "Ti-6-Al-4V Parameter High VED, Orientation 90°" showed good bio/cytocompatibility against both MC3T3 and L929 cell lines.