"Improving CT Reconstructions Through Regularization"

by Peter Cowal from Oregon State university

and  "Flash X-Ray Systems" by Gregory Detweiler from Oregon State University. 

in STAG 113

Link  to AMC seminar. 

Abstract of talk by Peter Cowal

The problem of computed tomographic (CT) reconstruction is ill-posed and benefits from regularization, especially when scan data is noisy or incomplete.  This talk will discuss historical and current regularization schemes applied to CT reconstruction, as well as their numerical implementation via the alternating direction method of multipliers (ADMM).

Abstract of talk by Gregory Detweiler

Conventional x-ray computed tomography (CT) attempts to image a snapshot of a static object in one instance of time. In fact, any movement of the object during the x-ray projection collection process degrades the quality of the CT reconstruction, and so attempts may be made to suppress object motion during the projection collection process, or methods for correcting for motion may be used after data collection. In contrast, in some cases one may actually desire to create a time series of CT reconstructions of an object undergoing some dynamic process. One method of doing so is to use a so-called flash x-ray system, where simultaneous x-ray projections from various angles can be collected several times during a dynamic event. The nature of this data acquisition process is severely limiting in the number of projections available for each CT reconstruction, leading to a severely sparse angular sampling CT reconstruction problem. In this talk I will introduce flash x-ray systems and the challenges inherent in CT reconstruction from data gathered from such systems and present an algorithm developed at Lawrence Livermore National Labs for performing such reconstructions.