NSF examines system for helping to match dental records
- By Joab Jackson
- May 26, 2004
SEATTLE'National Science Foundation research is developing a system to automate the matching of dental records to help narrow the search in identifying human remains.
Like the fingerprint, no two dental records are exactly the same, said Gamal Fahmy, a research assistant professor at West Virginia University.
The initial use for the technology will be to find the names of previously unidentified human remains. A set of teeth is the biometric that best stays intact after death, especially in extreme conditions such as during wartime, Fahmy said.
Fahmy and other researchers at Michigan State University and the University of Miami are developing what they call the Automated Dental Identification System, which is modeled after the Federal Bureau of Investigation's Automated Fingerprint Identification System. The FBI is providing test data for the project, as well as testing prototypes of the system.
Fahmy spoke at the fifth annual National Conference on Digital Government Research, held this week in Seattle. The conference is a forum for participants in NSF's Digital Government Research Program to share ideas and present research.
Fahmy said the Justice Department currently has a large database filled with information on unidentified corpses. It also has a database of missing people, and has dental records for many of them. Yet forensics investigators hoping to make a match between the two groups must match the dental records by hand, a time-consuming and nearly impossible task, given the large number of possible matches.
The application would sift through a database of possible matches, comparing them against the test subject. The closest matches will be presented to examiners, via a Web interface. A dental examiner must make the final identification.
The system currently uses MatLab mathematical software from The MathWorks Inc. of Natick, Mass., to visually outline the contours of each tooth in a set, which exist as X-rays. The system then uses MatLab and unique programs written in the C programming language to do pattern matching and to derive probabilities of how likely it is that two sets of teeth are the same.
Although the system now readily identifies individual teeth, the team is working to scale the system up to match entire sets of teeth, Fahmy said. It is also working on how to address the problems that fillings and missing teeth present to the matching process.
In addition to the Justice Department, the Defense Department could also use the application to help identify war casualties, Fahmy said.
The research team is developing the system with about $1 million in NSF funding, with additional funding from the National Institute of Justice.
Joab Jackson is the senior technology editor for Government Computer News.