System and method for hierarchical multi-level feature image synthesis and representation
A method for processing breast tissue image data includes processing the image data to generate a set of image slices collectively depicting the patient's breast; for each image slice, applying one or more filters associated with a plurality of multi-level feature modules, each configured to represent and recognize an assigned characteristic or feature of a high-dimensional object; generating at each multi-level feature module a feature map depicting regions of the image slice having the assigned feature; combining the feature maps generated from the plurality of multi-level feature modules into a combined image object map indicating a probability that the high-dimensional object is present at a particular location of the image slice; and creating a 2D synthesized image identifying one or more high-dimensional objects based at least in part on object maps generated for a plurality of image slices.
1. A method for processing breast tissue image data, comprising: processing image data of a patient's breast tissue to generate a set of image slices that collectively depict the patient's breast tissue;
applying one or more filters associated with a plurality of multi-level feature modules to each image slice of the set, wherein the multi-level feature modules are configured to recognize at least one assigned feature of a high-dimensional object that may be present in the patient's breast tissue;
at each multi-level feature module of the plurality, generating a feature map depicting regions in the respective image slice having the at least one assigned feature; and
combining the generated feature maps into an object map that indicates a probable location of the respective high-dimensional object, wherein the at least one feature of the high-dimensional object includes at least one of a low-level feature, a mid-level feature, and a high-level feature.
2. The method of claim 1 , further comprising creating a two-dimensional synthesized image of the patient's breast tissue, including identifying one or more high-dimensional objects based at least in part on object maps generated for a plurality of the image slices.
3. The method of claim 1 , wherein a learning library-based combiner is used to combine the generated feature maps into object maps.
4. The method of claim 1 , wherein a voting-based combiner is used to combine the generated feature maps into object maps.
5. The method of claim 1 , wherein the multi-level feature modules are configured to represent the respective at least one assigned feature of a high-dimensional image object.
6. The method of claim 5 , wherein the high-dimensional image object is defined to represent a specific breast lesion or a type of breast structure.
7. The method of claim 5 , wherein the high-dimensional image object comprises a three-dimensional image object.
8. The method of claim 1 , wherein the at least one feature of the high-dimensional object is the low-level feature, and wherein the low-level feature is based upon a general image filter selected from a group comprising an edge filter, a line filter, and a gradient filter.
9. The method of claim 1 , wherein the at least one feature of the high-dimensional object is the mid-level feature, and wherein the mid-level feature is based upon an algorithm or algorithmic model that represents and recognizes at least one of simple geometric shapes and image patterns.
10. The method of claim 9 , wherein the at least one of simple geometric shapes and image patterns comprise circular shapes, lobulated shapes and dense objects.
11. The method of claim 1 , wherein the at least one feature of the high-dimensional object is the high-level feature, and wherein the high-level feature is based upon an algorithm or algorithmic model that represents and recognizes at least one of complex geometric shapes and image patterns.
12. The method of claim 11 , wherein the at least one of complex geometric shapes and image patterns comprise types of breast lesions and breast structures.
13. The method of claim 1 , further comprising decomposing and representing the respective high-dimensional image object by a plurality of multi-level features designed to represent specific characteristics of the high-dimensional image object.
14. The method of claim 1 , further comprising generating a three-dimensional volumetric object map based upon respective object maps created for of a plurality of image slices.
15. The method of claim 14 , further comprising abstracting the three-dimensional volumetric object map to generate a three-dimensional volumetric object grid comprising a set of abstract three-dimensional image objects having differing attributes.
16. The method of claim 14 , wherein the three-dimensional volumetric object grid comprises object probability values for individual grid voxels, and wherein the differing attributes include one or more of location, size, and shape.
17. The method of claim 1 , wherein the object maps derived from a plurality of image slices are used to determine one or more of a location, size, shape, and morphology of the high-dimensional object.