1. Home
  2. /
  3. Research
  4. /
  5. Past project
  6. /
  7. Accuracy in the Comparison...
  • Thesis Title: Accuracy in the Comparison of Trabecular Bone Patterns and other Hard Tissue Features Visible on Digitised Dental Radiographs in Forensic Dental Identification
  • Author: Catherine Josephine McKenna
  • Submitted as a requirement for: Honours Degree of Bachelor of Science in Dentistry
  • Date: March 1998

Abstract: Significant trends in the reduction of dental caries and the increased use of dental restorative materials of varied radiopacity present problems for the forensic dental identification process. The potential for hard tissue features visible on dental radiographs to be used in the comparison process was investigated. Sets of three periapical views were taken of the posterior region of dry human mandibles with a vertical angulation difference of 10 degrees between each view.

The radiographs were converted to digital format using a charge-coupled device linear photodiode array slide scanner in 8-bit grey mode with a spatial resolution of 727×727 and contrast resolution of 256 grey levels. A graphic image software programme was used to establish an observer test for matching of fifty simulated ante-mortem and post-mortem images.

The images selected allowed comparison of hard tissue features only and included 25 true matches and 25 false matches. Image enhancement was performed on 22 of 50 matching sets. The observers gave a response on a scale rating of 1-4. Fourteen observers of varying background including five with forensic experience took part. The Kappa statistic was chosen to determine inter-rater agreement. Considerable variability was expected for responses to true matches with introduced geometric variation due to differences in angulation.

A higher rate of false positive responses was anticipated from observers with no forensic experience. Identical images were expected to illicit high positive responses from all observers. The same images were used to perform qualititative interpretation by digital subtraction and also for investigation of the grey level distribution across equivalent areas in both the “ante-mortem” and “post-mortem” images.