Compressed Sensing Based Computed Tomography Image Reconstruction

Abstract

In computed tomography (CT), an important objective is to reduce radiation dose without degrading image quality. The radiation exposure from CT scan will make severe problem in humans. This has high risk in the case of children and female. The higher exposure will lead to leukemia, cancer etc. So that low dose CT image reconstruction is the main concern now days. We have to reconstruct the image which gives better image quality from limited number of projection. Compressed sensing enables the radiation dose to be reduced by producing diagnostic images from a limited no of projections. According to compressed sensing theory the signal or image  can be reconstructed from the fewer samples and the sparse representation is the main objective behind it. The images are not sparse in nature, so some sparsifying transform is used for make the image to sparse. The object to be reconstructed scanned under sensors and several forward projections are takes place. In low dose CT we consider only smaller number of projections. From these projections the images are reconstructed. The CT image reconstruction is an ill-posed problem. That means solving underdetermined system of equations. This system solve the reconstruction problem using compressed sensing. This system chooses the noiselet as measurement matrix and haar wavelet as representation basis. The incoherence between measurement matrix and the representation basis is the one main property of compressive sensing. This incoherence will make the image reconstruction.