Patient-Specific Depth Error Correction in Dual-Sensor System for Image-Guided Minimally Invasive Facial Osteotomy

Abstract

This study introduces a patient-specific method for real-time depth error correction, improving the accuracy of the dual-sensor system previously developed by our team for image-guided facial osteotomy. The system integrates two RGB-D sensors with time-of-flight (ToF) technology and an optical localizer, enabling simultaneous real-time tracking of both the patient and surgical tools. Due to anatomical variations in facial skin, ToF sensors are susceptible to depth inaccuracies, requiring a tailored correction strategy. The proposed method uses the probe’s tip position, tracked by the optical localizer, to estimate parameters for an error model and correct depth inaccuracies. This real-time, patient-specific depth error correction significantly improves both registration and shape sensing accuracy. Validations using three patient mockups demonstrated a 87% reduction in depth errors, with an average fiducial registration accuracy of 2.33±0.23mm and an average shape sensing accuracy of 2.25±0.38mm. These results highlight the potential of patient-specific depth error correction to improve registration accuracy, thereby ensuring reliable and precise guidance, which may improve clinical outcomes in minimally invasive facial osteotomies.Clinical relevance—This method has the potential to enhance the accuracy of real-time guidance in image-guided facial osteotomies by correcting depth errors and improving image-to-patient registration accuracy, thereby contributing to better clinical outcomes and increased patient safety.