Research on improved image reconstruction algorithms of OCT based on incomplete projection data

Abstract

In order to improve the quality and solve the problem of low speed of image reconstruction in the traditional optical computerized tomography (OCT) when the data acquired is incomplete projection, the multiple constrained of genetic algorithm based on algebraic iterative was proposed. Generally speaking, under the condition of multiple-objective optimization, the common extreme point for all the objective functions doesn't exist. So we can achieve the preferable compromise in the contradictions of multiple objectives. In this article, there are three constrained conditions. The first one is the maximum entropy criterion which is used mostly to solve the problem of OCT image reconstruction when the data acquired is incomplete projection recently. The second one is the minimum criteria of peak value which is introduced to suppress noise effectively and ensure the gliding property of the image reconstruction, because of the first one leading to noise amplification during the iterative process. The last constrained condition is the minimum criteria of the difference between the projection again of image reconstruction and the original projection. The concept of penalize-function is introduced into the genetic algorithm, which would transform the constrained optimization problem to unconstrained. It is clearly demonstrated from the experiment results that the algorithm reconstruction technique can efficiently improve the quality of images reconstruction of the incomplete projection data.