Nonlinear partial volume artifact correction in helical CT

Partial volume (PV) artifacts occur when a scanned object is partially intruded into the scan plane. Because of the nonlinear effect of the PV, computed tomography (CT) images become unsuitable for diagnosis due to severe shading and streaking artifacts. In this paper, we present an iterative correction algorithm to combat these artifacts in helical scans. The algorithm first identifies potential partial volume objects by examining the object gradient based on multiple half-scan reconstructed images. The projection error is then estimated using a quadratic model. The final image is obtained by removing the error image (reconstructed with the projection error) from the original image. Various experiments were conducted to investigate the robustness and effectiveness of the proposed correction.