Method for identifying a material boundary in volumetric image data

A device for processing volumetric image data for identifying a material boundary within the data, the device comprising:

• a memory that stores a plurality of instructions;
• and
• a processor coupled to the memory and configured to execute the plurality of instructions to:
• obtain volumetric image data representative of anatomical region of a subject;
• for each of at least a subset of voxels in the anatomical region:
• identify a volumetric sub-region) surrounding the voxel, the sub-region including at least a subset of directly neighboring voxels,
• retrieve a pre-determined boundary transition function, the boundary transition function representing a model of expected voxel values as a function of distance across the material boundary to be identified,
• fit the pre-determined boundary transition function to the voxel values comprised within the volumetric sub-region, the fitting comprising determining fitting parameters of the model,
• determine a fit quality parameter indicative of a quality of the model function fit;
• identify candidate spatial points for the material boundary based on the determined fitting parameters and fit quality parameter for each voxel;
• estimate the material boundary within the data based on at least a subset of the identified candidate spatial points, wherein the boundary transition function models the voxel values as progressing from a minimum voxel value on one side of the material boundary to a maximum voxel value on the other side of the material boundary, and wherein the fitting parameters include at least one of:
• the minimum voxel value;
• the maximum voxel value;
• a perpendicular distance of the voxel from the boundary;
• and
• a unit normal vector of the boundary.