A method for 3-D volume image reconstruction of a subject. 2-D projection images are obtained over a range of scan angles, and one or more truncated images are identified from the obtained images. From each of the one or more truncated images, a corrected truncated image is formed by a repeated process of identifying a row of image data values in the one or more truncated images, assigning an extension length to one or both ends of the row according to the row location in the truncated image, and extending the truncated image data for the row of image data values by applying a row-dependent cosine function to each value in the extension length. A 3-D volume image is reconstructed using one or more of the corrected truncated images.