Projection Correlation Optimization using Simulated Annealing in CT Image Reconstruction

Abstract

Image reconstruction without analytical continuation is possible if and only if all projections cover entire object. In medical CT (Computed Tomography) applications, this condition is clearly not achievable. The new algorithm for image reconstruction using a few beam positions (projection correlation method) has been proved offer several advantages. . One drawback of using a few beam positions is a low quality image yielded. Thus, an objective function to rank the image quality is required to automate the decision process in determining the optimum beam positions. In this work, two optimization methods, gradient-descent method and simulated annealing, using projection correlation value as the objective function, were developed and compared to obtain the optimum beam positions. The effectiveness and practicality of each of these methods were compared. The results show that the simulated annealing presents superior results. This method could potentially be used in a clinical environment.