Effects of Different Reconstruction Kernels on Noise and Spatial Resolution of Computed Tomography Images: A Phantom Study

Noise and spatial resolution (SR) are the main characteristics of image quality in CT scanning affected by different radiation and technical parameters. One of the important parameters are reconstruction kernels. Different reconstruction kernels are used for noise reduction and/or edge-enhancement purposes. In this study, we investigate some reconstruction kernels about noise and SR on a commercial CT scanner.
Methods and Materials: To evaluate the noise and SR, a standard quality control phantom was used with the diameter of 16cm. Images were taken and reconstructed with 5 different kernels. Reconstructed kernels were H10, H30, H40, H50 and H70. For SR estimation we used point spread function (PSF) around narrow tungsten wire and full width at half maximum (FWHM) for each image. All measurements were performed on Siemens Emotion 6 CT scanner.
Noise of images in H10, H30, H40, H50 and H70 were 2.61±0.67, 4.33±1.20, 5.94±1.89, 8.58±3.30 and 19.59±4.53 respectively. FWHM in H10, H30, H40, H50 and H70 were 1.80±0.19, 1.44±0.11, 1.32±0.17, 1.24±0.09 and 1.04±0.17 respectively.
Discussion and In this study, the effect of implementing 5 different reconstruction kernels on noise and SR of CT images were investigated. Based on the results, the noise increased and FWHM decreased with the increment of kernels’ number from H10 to H70.