Fast reconstruction algorithm dealing with tomography artifacts

Abstract

Highly brilliant X-rays delivered by third generation synchrotron facilities coupled with modern detector technology permit routinely acquisition of high resolution tomograms in few minutes, making high throughput experiments a reality and bringing real-time tomography closer. New solutions for fast post-processing of such large amount of data are mandatory to fully exploit advantages provided by the high acquisition speed enabling new experiments until recently even unimaginable. The TOMCAT beamline¹ is well equipped for fast and high throughput experiments^{2, 3}. Here, we will focus on our solutions regarding the reconstruction process and discuss a fast reconstruction algorithm⁴, based on the Fourier Transform method as opposed to slower standard Filtered Back-Projection routines. We perform the critical step of such method, the polar-to-Cartesian mapping in the Fourier space, by convolution with the Fourier transform of functions with particular characteristics. This convolution approach combines speed with accuracy, making real-time data postprocessing closer to reality. This fast reconstruction algorithm implemented at TOMCAT also features several plug-ins, aimed at taming reconstruction artifacts. Here, we will discuss a new approach for removing rings from reconstructed datasets arising from defective detector pixels and/or damaged scintillator screens. This new method is based on a combined wavelet- FFT decomposition⁵. Another important feature of the presented reconstruction algorithm deals with local tomographic datasets, characterized by incomplete data. We show here that ad-hoc padding of the sinograms prior to reconstruction significantly reduces typical artifacts related to data incompleteness, making local tomography a valuable acquisition mode when small volumes in relatively large samples are of interest.