Impact of view reduction in CT on radiation dose for patients

Abstract

Iterative methods have become a hot topic of research in computed tomography (CT) imaging because of their capacity to resolve the reconstruction problem from a limited number of projections. This allows the reduction of radiation exposure on patients during the data acquisition. The reconstruction time and the high radiation dose imposed on patients are the two major drawbacks in CT. To solve them effectively we adapted the method for sparse linear equations and sparse least squares (LSQR) with soft threshold filtering (STF) and the fast iterative shrinkage-thresholding algorithm (FISTA) to computed tomography reconstruction. The feasibility of the proposed methods are demonstrated numerically.

CT image reconstruction with SuiteSparseQR factorization package

“…The Least Squares QR method, presented in (Paige and Saunders, 1982), is a good choice for sparse and possibly rank-deficient systems as is our case. We verified its validity in our previous works (Flores et al, 2014(Flores et al, , 2015Parcero et al, 2017;Chillarón et al, 2017).…”

“…On the other hand we find algebraic methods, that can be either iterative or direct. Iterative techniques such as ART (Andersen, 1989), SART (Andersen and Kak, 1984) or LSQR (Flores et al, 2014(Flores et al, , 2015Parcero et al, 2017) perform an approximation to the solution image and they obtain a good balance between radiation dose and image quality since they allow us reducing the numbers of projections taken when doing a CT scan. Nevertheless, they are really slow methods since they iterate to approximate the result image.…”

Evaluation of image filters for their integration with LSQR computerized tomography reconstruction method

“…That is why the algebraic methods of approximation started to be applied to reconstruct CT images, such as the LSQR (Least Squares QR) algorithm. They are capable of working with fewer views, as we have shown in our previous works [2][3][4][5]. But working with less projections also means more noise in the images.…”

“…N denotes the resolution of the image (128×128 pixels, 256×256 pixels, etc). This phase has been analyzed in previous works [2][3][4][5]9], using several different methods to solve the equations system. Nevertheless, in those works, the approach did not include an image filter combined with the iterative process of resolution, which we will study in this paper.…”

“…Also, it can be combined with a filter that eliminates Gaussian and Speckle noise, in our case we will use the Bilateral filter. And finally, it can be combined with the FISTA [16] acceleration technique which, as we have seen in [5], considerably reduces the number of iterations. The pseudocode of this process is shown in Algorithm 1, where we can specify which of the additional steps apart from LSQR we want to perform.…”