3D Reconstructions from Few Projections in Oral Radiology

Ralf K.W. Schulze (Klinikum der Johannes Gutenberg-University, Germany)
DOI: 10.4018/978-1-60566-292-3.ch007

Abstract

Established techniques for three-dimensional radiographic reconstruction such as computed tomography (CT) or, more recently cone beam computed tomography (CBCT) require an extensive set of measurements/ projections from all around an object under study. The x-ray dose for the patient is rather high. Cutting down the number of projections drastically yields a mathematically challenging reconstruction problem. Few-view 3D reconstruction techniques commonly known as “tomosynthetic reconstructions” have gained increasing interest with recent advances in detector and information technology.
Chapter Preview
Top

Radiographic Projection And 3D Reconstruction From Multiple Projections

The projection value measured by any x-ray sensitive receptor follows the well-known Lambert-Beer law:

(1) with I1 defining the intensity behind an absorber and I0 the input intensity, respectively. The parameters μ and d denote the mass absorption coefficient and the thickness of the absorber. We aim to estimate μ as shade of gray at discrete instances to obtain a reliable representation of the object.

In 1917, the Austrian mathematician Johann Radon discovered that the two-dimensional (2D) distribution of properties of an object may be obtained from an infinite number of line integrals sampled through the object. Mathematically, a function can be completely described by the complete number of straght line integrals through the support of, i.e.

(2)

The famous method predominantly applied for image reconstructions in CT-scanners uses this formula in a process termed “Filtered Backprojection (FBP)”. Assuming parallel x-rays, the relationship between the projection data (P) and the object are given by:

(3) where Θ denotes the projection angle and t the detector position in the beam (Fig. 1). Diracs delta function δ is required to define the line interval. The CT image reconstruction problem is to compute given . Note, that the term “” in equation (3)” represents a line equation (e.g. green “ray-line” in Fig. 1), (Beyerer & Leon, 2002) the sum of which the integrals are sampled. In other words, the measured data on the image receptor represent the integrals over a finite number of lines connecting the x-ray source with the detector cells.

Complete Chapter List

Search this Book:
Reset
Foreword
Petros Koidis
Preface
Acknowledgment
Chapter 1
Demetrios J. Halazonetis
\$37.50
Chapter 2
Jorg Hendricks, Gert Wollny, Alexander Hemprich, Thomas Hierl
\$37.50
Chapter 3
Antheunis Versluis, Daranee Tantbirojn
\$37.50
Chapter 4
Andreas Vogel
\$37.50
Chapter 5
Thomas Hierl, Heike Huempfner-Hierl, Daniel Kruber, Thomas Gaebler, Alexander Hemprich, Gert Wollny
\$37.50
Chapter 6
N.A. Borghese, I. Frosio
\$37.50
\$37.50
Chapter 8
Shital Patel, Yos Morsi
\$37.50
Chapter 9
Wei-Bang Chen, Chengcui Zhang
\$37.50
Chapter 10
Michele Jacotti, Domenico Ciambrone
\$37.50
Chapter 11
Ferenc Pongracz
\$37.50
Chapter 12
Antonios Zampelis, George Tsamasphyros
\$37.50
Chapter 13
Amit Chattopadhyay, Tiago Coelho de Souza, Oscar Arevalo
\$37.50
Chapter 14
Maxim Kolesnikov, Arnold D. Steinberg, Milos Zefran
\$37.50
Chapter 15
Anka Letic-Gavrilovic
\$37.50
Chapter 16
Petros Koidis, Marianthi Manda
\$37.50
Chapter 17
Hiroo Tamagawa, Hideaki Amano, Naoji Hayashi, Yasuyuki Hirose
\$37.50
Chapter 18
Nikos Nikolaidis, Ioannis Marras, Georgios Mikrogeorgis, Kleoniki Lyroudia, Ioannis Pitas
\$37.50