Studies On The Analytic Reconstruction Algorithm For Spiral Cone-beam CT

Spiral Cone-beam CT is superior to the traditional two-dimensional CT in datacollection, longitudinal resolution and rays'utilization ratio. Nowadays, it is more widelyused in medicine and industry CT. In this paper, we focus on the FDK and Katsevichalgorithm to do some deeply studies for helical cone-beam CT and the main context is asfollows:First, according to the study about the filter function of FDK approximationreconstruction algorithm for cone-beam CT, we proposed a tunable filter with exponentialwindow , which then was analyzed in the time domain and frequency domain, respectively.The experiments show that the error of the reconstructed results using the proposed filter issmaller than the common R-L and S-L filters, and the noise also can be reduced.Second, after described the Katsevich accurat reconstruction algorithms for spiral CT,we deduced the executed steps of Katsevich algorithm in details, which then was used toreconstruct the actual objects. According to analyze the artifact appeared when the endpointsof the PI-line were not treated appropriately, artifact was suppress by using the appropriateweighting function. Compared the experiments of the FDK and Katsevich algorithm, theKatsevich is better than the FDK without noise, while the FDK is better than Katsevich whenthe noise exist. This experimental results to appear is because the Katsevich algorithm needsto compute the derivation of projection data and it is sensitive to noise. To solve the problem,interpolation polynomial derivation method be used to compute the derivative of theprojection data. The experimental results show that this method makes the reconstructionresults more clearly.