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 a patient's breast tissue;
processing each image slice of the set of image slices to detect a first level feature of a high dimensional object present in the patient's breast tissue using an associated first-level feature module;
generating, for each image slice, a first-level feature map depicting zero or more regions having the first-level feature;
processing each image slice of the set of image slices to detect a second level feature of the high dimensional object present in the patient's breast tissue using an associated second-level feature module, wherein the first-level feature is a low-level feature and the second-level feature is a high-level feature;
generating, for each image slice, a second-level feature map depicting zero or more regions having the second-level feature; and
combining, for each image slice, the first-level feature map and the second-level feature map into an object map that indicates a probability region of the high-dimensional object in each image slice.
2. The method of claim 1 , wherein the first-level feature module is configured to recognize the first-level feature by applying one or more first-level recognition models; and
wherein the second-level feature module is configured to recognize the second-level feature by applying one or more second-level recognition models.
3. The method of claim 1 , wherein the first-level feature module is configured to recognize the first-level feature by applying one or more first-level recognition templates; and
wherein the second-level feature module is configured to recognize the second-level feature by applying one or more second-level recognition templates.
4. The method of claim 1 , wherein the first-level feature module is configured to recognize the first-level feature by applying one or more first-level recognition filters; and
wherein the second-level feature module is configured to recognize the second-level feature by applying one or more second-level recognition filters.
5. The method of claim 1 , wherein the first-level feature module is configured to recognize the first-level feature by applying one or more first-level recognition filters; and
wherein the second-level feature module is configured to recognize the second-level feature by applying one or more second-level recognition models.
6. The method of claim 1 , further comprising creating a two-dimensional synthesized image of the patient's breast tissue, including identifying the high-dimensional object based at least in part on the object maps generated for each of the image slices.
7. The method of claim 1 , wherein combining the first-level feature map and the second-level feature map into an object map comprises combining, by a learning library-based combiner, the first-level feature map and the second-level feature map into the object map.
8. The method of claim 7 , further comprising assigning a first weight to the first-level feature map and a second weight to the second-level feature map.
9. The method of claim 8 , further comprising adjusting at least one of the first weight and the second weight.
10. The method of claim 1 , further comprising applying a third filter associated with a third-level feature module to each image slice of the set, the third-level feature module configured to recognize a third-level feature of a high dimensional object present in the patient's breast tissue; and
generating, for each image slice, a third-level feature map depicting zero or more regions having the third-level feature.
11. The method of claim 10 , wherein the third-level feature is different from both the first-level feature and the second-level feature.
12. The method of claim 11 , wherein the third-level feature is a mid-level feature.
13. The method of claim 12 , wherein the probability region indicates one or more of a location, a size, and a scope of the high-dimensional object.
14. The method of claim 1 , wherein the probability region comprises a probability gradient.
15. A system comprising:
a non-transitory computer-readable memory storing executable instructions; and
one or more processors in communication with the computer-readable memory, wherein, when the one or more processors execute the executable instructions, the one or more processors perform:
processing image data of a patient's breast tissue to generate a set of image slices that collectively depict a patient's breast tissue;
processing each image slice of the set of image slices to detect a first level feature of a high dimensional object present in the patient's breast tissue using an associated first-level feature module;
generating, for each image slice, a first-level feature map depicting zero or more regions having the first-level feature;
processing each image slice of the set of image slices to detect a second level feature of the high dimensional object present in the patient's breast tissue using an associated second-level feature module, wherein the first-level feature is a low-level feature and the second-level feature is a high-level feature;
generating, for each image slice, a second-level feature map depicting zero or more regions having the second-level feature; and
combining, for each image slice, the first-level feature map and the second-level feature map into an object map that indicates a probability region of the high-dimensional object in each image slice.
16. A non-transitory computer readable medium having stored theron one or more sequences of instructions for causing one or more processors to perform:
processing image data of a patient's breast tissue to generate a set of image slices that collectively depict a patient's breast tissue;
processing each image slice of the set of image slices to detect a first level feature of a high dimensional object present in the patient's breast tissue using an associated first-level feature module;
generating, for each image slice, a first-level feature map depicting zero or more regions having the first-level feature;
processing each image slice of the set of image slices to detect a second level feature of the high dimensional object present in the patient's breast tissue using an associated second-level feature module, wherein the first-level feature is a low-level feature and the second-level feature is a high-level feature;
generating, for each image slice, a second-level feature map depicting zero or more regions having the second-level feature; and
combining, for each image slice, the first-level feature map and the second-level feature map into an object map that indicates a probability region of the high-dimensional object in each image slice.