Effective Estimation of Focal Spot Isocenter for Sinograms Obtained with Flying Focal Spot Scans for Filtered Back Projection
Y Yu1*, B Whiting2, D Politte3, J Williamson4, (1) VCU, Richmond, VA, (2) Mallinckrodt Inst of Radiology, PITTSBURGH, AA, (3) Washington Univ, Saint Louis, MO, (4) Virginia Commonwealth University, Richmond, VASU-E-I-23 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall
Purpose: To validate the determination of parameters required to preprocess sinograms obtained by fan beam CT with flying focal spot (FFS) scans, such that a combined data set could be used with existing filtered back projection (FBP) reconstruction software based on single focal spot (SFS) geometry.
Methods: The preprocessing step requires determination of the isocenter for each FFS in the target reconstruction geometry. A Philips Brilliance Big Bore CT was used to obtain FFS sinograms, which double samples the number of gantry positions relative to SFS sinograms, having the same number of detector measurements for each gantry position. Mass weighting of sinograms of a Teflon rod scan and curve fitting were used to determine isocenters for focal spots. FFS sinograms were reformatted before FBP such that each pair of adjacent gantry-step measurements were combined by interleaving detector readings. The interleaved measurements were assigned to the average gantry angle. Image transition sharpness was compared for 33 images of the Teflon rod reconstructed with 1/8 focal spot deflection difference to test the isocenter result. SFS and FFS Sinograms of a high-resolution bar pattern phantom were reconstructed by an in-house reconstruction code package for resolution comparison. In-house reconstructed images of the Teflon rod were compared with scanner images for verifying the effectiveness of this sinogram reformatting method.
Results: Image profile was the sharpest when the Teflon rod sinogram was reconstructed with isocenter parameters found by the curve fitting experiment. It was comparable to the sharpness of scanner image profile when the in-house image was reconstructed with resolution of 1.0mm FWHM.
Conclusions: The good clinical resolution of images reconstructed with investigated FFS isocenters suggests that the experiment of estimating focal spot centers and reformatting sinograms is effective for utilizing reconstruction software that does not support FFS geometry.