Abstract:
HTML
Purpose: Polychromatic x-rays and scattered radiation lead to an underestimation of object attenuation during measurement. This underestimation is more pronounced for regions of high attenuation lengths than for regions of low attenuation lengths. The resulting images show cupping artifacts with darker areas in the center of the object than in the periphery. Since the physical effects of scatter cannot be predicted in acceptable time we propose an empirical approach to determine a precorrection function that removes the cupping artifact.
Methods and Materials: Our approach is based on CT projection data of homogeneous water-equivalent phantoms of varying size with circular and oval shape. These are used to calibrate a set of correction coefficients. For each phantom a CT reconstruction is performed. In our case this is a Feldkamp reconstruction for circular trajectories. The reconstructed volume exhibits the said cupping artifacts. It is thresholded using a threshold of T=-500 HU to separate between water and air. Voxels below T are set to 0 HU (air) and those above T are set to 0 HU (water). The resulting binary volume is reprojected using the same ray geometry as the original measurement to obtain ideal projection data. A least square minimizing fit using a quadratic precorrection function pideal=a p2+b p+c where p is the measured attenuation value was found to be sufficient to locally describe the discrepancies between the physical data and the ideal data. We allowed the coefficients a, b, and c to vary slowly as a function of the detector coordinates u and v to include varying scatter contribution. ECC was evaluated using simulations and measurements of a micro CT scanner (Tomoscope 10010M, VAMP GmbH, Möhrendorf, Germany) and a rotating C-arm unit equipped with a flat panel detector (Axiom Artis, Siemens Medical Solutions, Forchheim, Germany). The objects scanned with the C-arm were an abdomen and a thorax phantom with various extension rings. With the micro CT a kimberlite core was scanned and several mice were imaged in vivo.
Results: Reconstructions of homogeneous objects showed a reduction of the cupping amplitude from typically 50 to 200 HU to about 0 to 5 HU (no cupping). Cupping artifacts were effectively removed in all simulated and measured scans. The best results were obtained when the reconstructed objects is similar in size and shape to the phantom used for calibration.
Conclusion: ECC is an efficient tool to reduce cupping artifacts from cone-beam CT data. Based on a calibration scan it reduces the combined effects of beam hardening and scatter.
Questions about this event email: marc.kachelriess@imp.uni-erlangen.de
Kachelriess PhD, M,
Empirical Cupping Correction (ECC) for X-Ray Cone-Beam CT. Radiological Society of North America 2003 Scientific Assembly and Annual Meeting, November 30 - December 5, 2003 ,Chicago IL.
http://archive.rsna.org/2003/3106849.html
