Hybrid Fiducial System for TunedAperture Computed Tomography

1
Hybrid Fiducial System for Tuned-Aperture
Computed Tomography

• R.L. Webber, S.B. Robinson, and F.H. Fahey
• Wake Forest University

Supported in part by NIH RO1 DE12227, DAMD
17-98-1-8349, and Instrumentarium Imaging
2
Objectives

• Simplify extraction of essential fiducial
  information
• Eliminate obscuration of desired information by
  superposition of fiducial shadows
• Facilitate quantitative 3D radiography with
  conventional (2D) imaging systems

3
Methods

• Use digital camera to isolate radiolucent
  fiducials from underlying tissues as seen from a
  variety of projection angles
• Geometrically couple resulting photographic image
  with respective simultaneously exposed x-ray
  projection images
• Use photographic data to infer the associated
  x-ray projection geometry required for 3D
  reconstruction
• Compute 3D radiograph using TACT algorithm

4
Projection Schematic
5
Mathematical Justification

• P(P1,P2,,Pn) are x-ray source positions
• P(P1,P2,,Pn) are corresponding camera
  positions
• x is point in object projected by x-ray beam
• x is point in object seen by coupled camera
• Theorem Whenever an image of a point xi is
  created using pi, then the backprojection using
  pi yields an unambiguous point x, i.e. the
  homologous projection lines intersect at x.
• This geometry can be realized with reasonable
  coupling constraints

6
Methods (contd)
Expose isolated tooth from 12 arbitrary angles
7
Methods (contd)
Photograph coupled fiducials from 12 homologous
angles
8
Results
Reconstruct arbitrary slices from optically
registered x-ray projections
9
Results (contd)
Reconstruct arbitrary 3D projections from same
data
10
Conclusions

• Hybrid photographic/radiographic system can
  produce accurately registered 3D reconstructions
  using TACT algorithms
• Potential obscuration of information caused by
  overlying fiducial shadows can be eliminated
  completely from TACT 3D reconstructions
• Hybrid method facilitates automated 3D analysis
  by simplifying isolation of essential fiducial
  registration data from radiographic information