Interpolation of CT Slices for 3-D Visualization by Maximum Intensity Projections

Abstract

Visualization of 3-D volume data through maximum intensity projections (MIP) requires isotropic voxels for generation of undistorted projected images. Unfortunately, due to the inherent scanning geometry, X-ray computed tomographic (CT) images are mostly axial images with submillimeter pixel resolution, with the slice spacing on the order of half to one centimeter. These axial images must be interpolated across the slices prior to the projection operation. The linear interpolation, due to the inherent noise in the data, generates MIP images with noise whose variance varies quadratically along the z-axis. Therefore, such MIP images often suffer from horizontal streaking artifacts, exactly at the position of the original slices (e.g., in coronal and sagittal MIPs). We propose a different interpolation technique based on a digital finite impulse response (FIR) filter. The proposed technique flattens the change in noise variances across the z-axis and results in either elimination or a reduction of horizontal streaking artifacts in coronal and sagittal views.