Super-resolution of medical volumes based on Principal Component Regression

In medical imaging, the data resolution is usually insufficient for accurate diagnosis in clinical medicine. Especially in most case, the resolution in the slice direction (Z direction) is much lower than that of the in-plane resolution (XY direction). Therefore it is difficult to construct isotropic voxels, which is very important in 3-D visualization systems, such as surgical system. In this paper, we propose a method for improving resolution in the slice direction for medical volume images based on Principal Component Regression (PCR), which can be considered as one of the learning based super-resolution techniques. The experimental results verify the effectiveness of the proposed method by comparison with the conventional interpolation methods.