Validation of the blurring of a small object on CT images calculated on the basis of three-dimensional spatial resolution.

Abstract

We determine three-dimensional (3D) blurring of a small object on computed tomography (CT) images calculated on the basis of 3D spatial resolution. The images were characterized by point spread function (PSF), line spread function (LSF) and slice sensitivity profile (SSP). In advance, we systematically arranged expressions in the model for the imaging system to calculate 3D images under various conditions of spatial resolution. As a small object, we made a blood vessel phantom in which the direction of the vessel was not parallel to either the xy scan-plane or the z-axis perpendicular to the scan-plane. Therefore, when scanning the phantom, non-sharpness must be induced in all axes of the image. To predict the image blurring of the phantom, 3D spatial resolution is essential. The LSF and SSP were measured on our scanner, and two-dimensional (2D) PSF in the scan-plane was derived from the LSF by solving an integral equation. We obtained 3D images by convoluting the 3D object-function of the phantom with both 2D PSF and SSP, corresponding to the 3D convolution. Calculated images showed good agreement with scanned images. Our technique of determining 3D blurring offers an accuracy advantage in 3D shape (size) and density measurements of small objects.