Self-Updating Neural Networks for Real-Time Canal Morphology Mapping During Rotary Instrumentation

Abstract

Accurate real-time mapping of root canal morphology is critical for safe and efficient rotary instrumentation in endodontic procedures. Traditional imaging methods often fail to provide dynamic feedback during instrumentation, limiting the precision of treatment. This study proposes a self-updating neural network framework capable of continuously learning from intraoperative sensor and imaging data to model canal morphology in real time. By integrating adaptive learning mechanisms, the network updates its predictions as new information becomes available, enabling precise guidance during rotary instrumentation. Preliminary evaluations demonstrate improved mapping accuracy and responsiveness compared to static models, highlighting the potential of AI-driven adaptive systems in enhancing endodontic outcomes. This approach represents a significant step toward intelligent, real-time dental navigation systems.