Ordered-Map Tree Structure for Non-Rigid Volume Registration

The patent segments the deformation map checking process by organizing deformation maps into a tree structure where only relevant maps need to be checked for each voxel. The tree partitions the search space so that boundary checking is performed only on deformation maps that actually affect each voxel, rather than exhaustively checking all maps. This segmentation maintains registration accuracy while dramatically reducing processing time.

The patent performs preliminary organization of deformation maps into a tree structure before the actual registration process. This preliminary action pre-computes and stores the hierarchical relationships and bounding boxes, so that during registration, the system can quickly navigate to only the relevant deformation maps for each voxel without performing exhaustive checks. The tree structure is built in advance to enable efficient traversal during on-the-fly registration.

The tree structure segments the deformation maps into hierarchical levels, allowing the system to quickly identify and check only the relevant maps for each voxel during on-the-fly registration. This segmentation ensures that even with reduced boundary checking, the accuracy is maintained by systematically determining which maps need to be checked based on the voxel's position and the tree's spatial organization.

The tree structure acts as an intermediary between the voxel and the deformation maps. Instead of directly checking all deformation maps for each voxel, the tree mediates by providing a structured pathway to identify only the relevant maps. This intermediary mechanism enables on-the-fly registration to proceed efficiently while maintaining accuracy through systematic selection of deformation maps based on spatial relationships stored in the tree.

The patent segments the deformation maps into a tree structure that organizes maps by spatial regions and priorities. This segmentation transforms a simple but inefficient linear search into a structured hierarchical search, where the complexity is increased in the data organization but the searching time is dramatically reduced by eliminating unnecessary checks of irrelevant deformation maps.

The tree structure is pre-computed and stored before the registration process, performing the complex organization work in advance. This preliminary action creates an optimized data structure that, while more complex to build, enables extremely fast traversal during actual registration. The searching time is reduced because the tree's hierarchical organization allows the system to skip large portions of the deformation map set that do not affect each voxel.