Aliasing artifact suppression with adaptive segmentation based edge enhancement

One of the inherent limitations of the third generation CT scanner is the projection undersampling. Because of the fan-beam geometry, patient motion, and many other factors, the Nyquist sampling criteria are not always strictly observed. As a result, the fine structures of the anatomy and important pathologies are often marred by aliasing streaks, which render the image unusable. We analyze the root cause of the aliasing artifact and present an adaptive algorithm that enhances the fine structures of the anatomy and suppresses aliasing artifacts and noise. The algorithm first reconstruct an image with a modified reconstruction kernel which preserves as much high frequency information as possible without introducing significant aliasing artifacts. The resulting image is then segmented into two classes. A fuzzy classification method is employed which uses not only the pixel intensity and texture information, but also the classifications of adjacent slices. Various phantom and clinical studies have demonstrated the robustness and effectiveness of our approach