An efficient low-dose CT reconstruction technique using partial derivatives based guided image filter

Low-dose Computed Tomography (CT) reconstruction techniques have been implemented to minimize the X-ray radiation in a human body. Many researchers have designed different low-dose CT reconstruction techniques to reduce the effect of radiation in a human body. However, the majority of these techniques suffer from over-smoothing, edge distortion, halo artifacts, gradient reversal artifacts etc. problems. Therefore, in this paper, novel partial differential equations and dictionary learning based reconstruction technique have been designed to reconstruct the low-dose CT images. Extensive experiments have been carried out to evaluate the effectiveness of the proposed technique that existing image reconstruction techniques. It has been observed that the proposed technique significantly preserves the radiometric information of low-dose CT images with a lesser number of edge distortion, halo and gradient reversal artifacts. Also, the proposed technique is computationally faster than existing techniques, therefore most suitable for real-time low-dose CT reconstruction systems.