A recent study at the University of Campinas has developed an artificial intelligence (AI) model that significantly improves CT scan quality for patients with dental implants, thereby enhancing diagnostic accuracy.
High-quality CT images are crucial for cases requiring in-depth visualization of oral and dental structures. However, metal implants in the mouth can interfere with X-rays, causing artifacts, streaks, or smears in the images, complicating diagnosis.
This innovative technology was developed by Matheus Lima Oliveira, Professor of Dental Radiology at the University of Campinas’ Piracicaba Dental School (FOP), as part of his postdoctoral research. This work was conducted in collaboration with the University of Basel, Switzerland, and was supported by the São Paulo Research Foundation (FAPESP) and the Swiss Society of Dento-Maxillofacial Radiology (SGDMFR).
The technology received the Artificial Intelligence and Technological Advancement Award at the 25th International Congress of Dento-Maxillofacial Radiology (IADMFR), held in London, UK, on June 24-25.
The Challenge of Metal Artifacts
Traditional CT scans produce three-dimensional structural images, providing doctors with more detailed diagnostic information than standard X-rays. However, when a patient has metal dental implants or braces in their mouth, the interaction between X-rays and the metal can cause “metal artifacts,” which can affect image quality.
This phenomenon stems from X-ray beam hardening—when radiation photons pass through metal, some are absorbed while others reach the sensor, resulting in uneven brightness or streaking in the image. Traditional methods often require adjusting the patient’s position or focusing the scan on specific areas, but completely eliminating artifacts remains difficult.
Innovative Application of AI Models
To address this challenge, Oliveira proposed developing an algorithm capable of identifying metal artifacts in images and restoring the original image. The research team used fresh pig jawbones for training and implanted dental implants made of three common metals: titanium, zirconium oxide, and titanium-zirconium alloy. CT scans were generated for approximately 400 combinations.
By comparing artifact-free and artifact-enhanced images, the AI system successfully identified artifacts and reconstructed images in 100% of tests. Notably, the model can also handle artifacts generated even when implants are located outside the scan’s region of interest.
Future Outlook
The integration of AI and dental imaging is expanding, showing great potential particularly in radiology. Oliveira stated that the current results are still preliminary, primarily in simulations, and require further validation in patients.
His ultimate goal is to develop a system that prevents artifacts during image generation, rather than simply correcting them afterward.
This research marks a significant advancement in AI for dental CT scanning, providing a new tool for improving diagnostic accuracy in patients with dental implants and opening up broad prospects for future clinical applications.
