The registration problem is one of the most important tasks to find the optimum spatial transformation between two sets of features. To reduce subtraction artifacts, we propose a new temporal subtraction method based on a voxel matching technique by warping the previous image to match the current one . Figure 4.8 shows the overall scheme of our proposed method for temporal subtraction. The main steps are (1) global matching, (2) local matching, (3) 3D nonlinear image warping based on voxel matching technique, and (4) 3D image subtraction steps.
In general, images generated at different patient visits can vary in voxel size. In this study, we adjusted the voxel size on the X- and Y-axes of a previous CT image to match that of a current CT image using linear interpolation. To segment the lung region as a VOI, we identified a lung region in the current CT image by applying a 3D Gaussian filter before the binarization technique based on the Otsu method  and a morphological filter.