A comparison of two registration techniques for digital subtraction radiography

S. M. Dunn, P. F. van der Stelt, A. Ponce, K. Fenesy, S. Shah

Research output: Contribution to journalArticle

37 Scopus citations

Abstract

The dental radiograph is a diagnostic tool that is used to view internal structures without any invasive procedures. A difference radiograph, i.e. one showing the changes that occurred over time, is the best tool for diagnosing slow or small changes. The difficulty in creating a meaningful difference radiograph, or its digital image, is that the imaging geometry of the image pair must be identical. This paper addresses the problem of reproducing imaging geometry. A new technique for registering digital images of radiographs based on the correspondence of 3D anatomical structures (such as root apices and the cemento-enamel junction) was compared with a registration method using occlusal stents. In a pilot study of 30 volunteers, difference images were created by three different registration techniques (no registration, mathematical methods and occlusal stents). The standard deviations of the difference images were used to compare the registration techniques. The results show that this new mathematical technique applied to digital images of radiographs can be used to establish correspondence between pairs of clinical images taken at different projection angles and to produce reconstructed images comparable with images taken with occlusal stents.

Original languageEnglish (US)
Pages (from-to)77-80
Number of pages4
JournalDentomaxillofacial Radiology
Volume22
Issue number2
DOIs
StatePublished - May 1993

All Science Journal Classification (ASJC) codes

  • Otorhinolaryngology
  • Radiology Nuclear Medicine and imaging
  • Dentistry(all)

Keywords

  • radiographic image enhancement
  • Radiography, dental
  • signal processing, computer assisted

Fingerprint Dive into the research topics of 'A comparison of two registration techniques for digital subtraction radiography'. Together they form a unique fingerprint.

  • Cite this