Comparison of reconstruction algorithms to extend CT reconstruction field-of-view

Abstract

This paper evaluates the impact of reconstruction algorithms to extend CT reconstruction field-of-view (DFOV) on radiation dose calculation for oncology treatment planning. Three recon algorithms for extending DFOV are compared: 1) a symmetric-mirroring (SM) algorithm; 2) a sinogram interpolation (SI) algorithm; and 3) a water-cylinder fitting (WF) algorithm. Each of the IQ and anthropomorphic phantoms is scanned twice using GE Lightspeed 16. The first scans, where the phantoms are positioned inside the 50 cm DFOV, are used to establish the "ground truth"; while in the 2nd scans, the phantoms are positioned in such way that the edges are 4-8 cm outside the 50 cm DFOV. For quantitative assessment, a point of interest within 50 cm DFOV is chosen for each phantom, to which the accumulative attenuation in all views is calculated as a lst-order approximation of dose to be delivered to the point. The performance of the algorithms is evaluated by accumulative attenuation accuracy by comparing with the "ground truth". In reconstruction from all three algorithms, objects outside 50 cm DFOV are restored but image quality and accumulative attenuation varies among algorithms. The SM reconstruction results in restoration of low-attenuating objects in small truncation cases, but its ability to restore high-attenuating objects in large truncation cases is limited and produces worst attenuation accuracy due to the overly simplicity of the algorithm. The SI algorithm provides improved accuracy even in large truncation cases, especially for large-attenuating objects. But the CT number still significantly depart from the ground truth due to the lack of consistency in sinogram interpolation. WF produces significantly better image quality and >17% higher accuracy than SM and SI in all cases. In summary, it is feasible to extend the CT DFOV with algorithmic corrections. Among the selected algorithms, the WF algorithm provides reconstructions with the best quality for oncology treatment planning.