A kernel of popcorn that went unpopped. A piece of ice like an iceberg. Grueling nighttime grinding. Who knows what started a fissure down the length of that tooth?
As the pain mounts, so does the frustration, because longitudinal tooth fractures -- which extend through the long axis of the tooth and expand with time -- are often invisible to the naked eye and difficult to detect with conventional radiology.
Such teeth can rarely be saved, so dentists and patients would both benefit from discovering a fracture right away. "It would prevent a lot of unnecessary treatment and suffering," says André Moll, DDS, MS, PhD, an Oral and Maxillofacial Radiologist and Assistant Professor of Diagnostic Science and General Dentistry at the University of North Carolina, Chapel Hill.
That’s why Moll and his colleagues were eager to try out local computed tomography (LCT) as a method of detecting these problem cracks. This new technology, which is a form of cone beam CT, produces high resolution 3D images of a small area of the jaws
As they reported in the June 2007 Oral Surgery,Oral Medicine, Oral Pathology, Oral Radiology, and Endodontology, the method works well -- at least when applied to extracted teeth under laboratory conditions.
In their experiment, the researchers anchored 30 whole teeth in acrylic, one by one, to prevent the roots from splitting, then coated them with wax for easy removal. Then they placed a screwdriver-style in the canals and gently tapped, fracturing the teeth from within the root. These 30 split teeth were matched to 30 whole teeth that were used as controls.
To simulate conditions within the mouth, they placed the teeth in an empty socket of an otherwise dentate mandible, and before scanning they surrounded the mandible with boxing wax, which simulated soft tissue. Silicone dental wax served as a surrogate for trabecular bone, and the researchers mixed in soy grains to simulate marrow spaces.
Then the scanning began. The investigators used a conventional cone-beam dental x-ray source, the Planmeca Prostyle Intra, and a Schick CDR detector, which has high-resolution digital capacity. The mandible sat on rotator and the detector matrix was aligned with the rotational axis of the teeth.
The source sat 65 cm from the tooth, and the detector sat 12.5 cm away. As each tooth was scanned it was placed in the center of rotation, with the long axis of the tooth parallel to the rotation axis. This allowed the system to take 180 images with 1 degree of separation in a 180-degree arc. This created 3D images, which the investigators hoped would offer an advantage over conventional 2D x-rays. Conventional x-rays usually can't show fractures unless the x-ray beam is parallel to the plane of the facture.
A filtered back projection reconstruction algorithm allowed the investigators to generate multiple axial slices with a thickness of 0.2 mm. They removed the top slices showing the crown to prevent observers from judging the fractures based on the status of the crowns, which were damaged in producing the fractures.
For comparison purposes, the researchers took standard 2D x-rays of each tooth using an Eastman Kodak RVG 6000 sensor. To likewise prevent the observers from being clued in, they masked the crowns with an opaque overlay.
Finally, the investigators put their images to the test, showing them to four radiologists, one radiology student, four endodontic residents, and a periodontist. Using three side views (orthogonal planes to view the image volume), the observers were asked to rate the probability on a scale of 1 to 5 that each tooth had a fracture.
The results suggest that LCT is significantly more accurate than traditional radiography. The investigators tested the observers’ ratings of the teeth by plotting their "false positive" responses against their "true positive" responses and measuring the area under the curve (a receiver operating characteristic curve). The observers scored an average of 0.91 with LCT vs. 0.70 with conventional radiography. That's equivalent to a 91 percent versus 70 percent accuracy rate, had the observers given simpler "yes or no" answers to the question of whether each tooth was fractured (a two-alternative forced choice comparison), Moll said.
The UNC researchers cautioned that the results don’t prove the new technology will work in a real-world situation with a human mouth. Real-life fractures may not resemble a tooth split with a screwdriver. Nor was there real bone or ligament surrounding the teeth in this experiment. Likewise, in actual practice, the teeth with longitudinal fractures dentists encounter may already be restored in some way, often with posts.
These differences between laboratory conditions and the real world may explain why the observers were able to identify so many of the fractured teeth using conventional radiography -- a finding that somewhat surprised the investigators.
Engineers must resolve another problem before the technology is truly practical: creating a 3D image from 180 images meant using 180 times as much radiation as in the standard x-ray, an unacceptably high dose.
"We've done some research since then to see if we could bring the dosage down," said Moll. "We were successful, but it's really a problem we need to work with our industry partners to solve."
Overall, he said, LCT has significant potential to help solve a problem that has bedeviled dentists since the profession began.