Polymerization Strategy for the Compression, Segmentation, and Modeling of Volumetric Data

We present a data structure for the representation of volumetric data. The data structure is designed to allow for easy compression, storage, segmentation, and reconstruction of volumetric data. We call our data structure the L-block, abstracting many of the properties of Lego® blocks, and refer to the process of creating and manipulating L-blocks as the polymerization strategy. The concept of an enhanced volume data set (EVDS) is introduced, where the data set is enhanced by explicitly introducing Boolean labeling of edges between adjacent voxels of the volume data. This enhancement, by “polymerizing” adjacent connected voxels into connected components, facilitates real-time data compression and segmentation of embedded objects within the volume data set. These connected components are packaged in the new container type, the L-block, with the intention of efficiently packaging the connected components with a minimum of adjacent unconnected voxels. We present the L-block data structure in detail. We describe methods for compressing volume data using the L-block structure, intersecting and merging L-blocks, and segmenting data. While the L-block data structure is general, it was developed to represent scanned brain microstructure at a neuronal level of detail. We highlight the performance of our implementation of the polymerization strategy on a set of sampled neuronal data.