AI-driven automatic segmentation of oral surgery-related tissues in cone beam computed tomography images may offer more efficient dental scanning, according to a new study.
The research, published in the International Journal of Oral Science, evaluated a fully automated system that was found to enhance the precision and efficiency of dental implant surgeries by significantly improving tissue segmentation speed and accuracy.
Computed tomography (CBCT) technology plays a "crucial" role in modern digital dentistry by providing high-resolution 3D images for detailed assessments of oral anatomy, noted the researchers. The precision is essential for meticulously planning implant positions to avoid damaging vital structures like nerves and blood vessels.
The researchers streamed the process through an advanced algorithm that automatically segmented alveolar bone, teeth, and maxillary sinus with high accuracy. They reported Dice scores—a statistical measure of similarity between two sets—reaching 96.5% for teeth, 95.4% for alveolar bone, 93.6% for maxillary sinus, and 94.8% for the mandibular canal indicating superior performance of the system, outperforming previous methods.
This approach also includes an image preprocessing method tailored to adapt to different CBCT machine outputs by analyzing data distribution histograms, the researchers noted. This addresses the common challenge of variations in image quality due to different imaging protocols across CBCT devices.
The deep learning methods were based on convolutional neural networks (CNNs). The study demonstrated advancements in handling the complexities of segmentation tasks. The study also introduced automated identification of tooth positions based on the FDI Two-Digit Notation essential for “the correct localization and assessment of the patient’s oral health status," researchers noted.
By reducing reliance on manual segmentation, the method significantly enhanced the efficiency and accuracy of surgical planning in digital dentistry, particularly in implant surgery. The system's ability to automatically and accurately map dental structures directly from CBCT scans helps accelerate the development of digital dentistry, according to the researchers. This automation not only reduces the time taken but also minimizes potential errors associated with manual processes, leading to more reliable and expedited patient care.
The researchers declared no competing interests.
