High-throughput biomarker segmentation utilizing hierarchical normalized cuts
A method and apparatus for obtaining segmented images of the stained regions may comprise quantifying the extent of the presence of staining of a biomarker in an original image of a sample, which may comprise selecting a domain swatch of data based upon a user specified domain knowledge; clustering the data within the original image by conducting a frequency weighted mean shift of the data within the original image to convergence, forming a hierarchical plurality of layers each having a different data resolution to form a hierarchical data pyramid; segmenting the plurality of mean shifted data images to determine in each mean shifted data image within the hierarchical data pyramid data not excluded as outside of the swatch; mapping the data not excluded as outside the swatch spatially back to the original image to create a final image; and, storing the final image on a storage medium for further analysis.
1. A high-throughput method for quantifying staining of a biomarker in an original image of a pathology sample comprising the steps of:
a. selecting a domain swatch of data based upon a user specified domain knowledge of data within the original image;
b. clustering the data within the original image by conducting a frequency weighted mean shift of the data within the original image to convergence to form a hierarchical plurality of layers of mean shifted data images, each having a different data resolution to form a hierarchical data pyramid;
c. segmenting the plurality of mean shifted data images to determine in each mean shifted data image within the hierarchical data pyramid data included within the swatch;
d. mapping the data included within the swatch spatially back to the original image to create a final image; and
e. storing the final image on a storage medium for further analysis.
2. The method of claim 1 , wherein the pathology sample comprises an image of a tissue cylinder from within a tissue micro-array.
3. The method of claim 1 , wherein the pathology sample comprises a tissue biopsy sample.
4. The method of claim 1 , wherein the biomarker is a vascular biomarker.
5. The method of claim 1 , wherein the biomarker is stained for indicating a disease condition.
6. The method of claim 1 , wherein the domain swatch of data within the original image is selected from the group consisting of image color, intensity and texture.
7. The method of claim 1 , wherein said segmentation step comprises a normalized cuts algorithm.
8. A high-throughput method for detecting regions of interest due to presence of a biomarker in an original image of a pathology sample comprising the steps of:
a. selecting a domain swatch of data based upon a user specified domain knowledge of data within the original image;
b. clustering the data within the original image by conducting a frequency weighted mean shift of the data within the original image to convergence to form a hierarchical plurality of layers of mean shifted data images, each having a different data resolution to form a hierarchical data pyramid;
c. segmenting the plurality of mean shifted data images to determine in each mean shifted data image within the hierarchical data pyramid data included within the swatch;
d. mapping the data included within the swatch spatially back to the original image to create a final image;
e. storing the final image on a storage medium for further analysis; and
f. detecting regions of interest on the final image.
9. The method of claim 8 , wherein the pathology sample comprises an image of a tissue cylinder from within a tissue micro-array.
10. The method of claim 8 , wherein the pathology sample comprises a tissue biopsy sample.
11. The method of claim 8 , wherein the biomarker is a vascular biomarker.
12. The method of claim 8 , wherein the biomarker is stained for indicating a disease condition.
13. The method of claim 8 , wherein the domain swatch of data within the original image is selected from the group consisting of image color, intensity and texture.
14. The method of claim 8 , wherein said segmentation step is via the normalized cuts algorithm.
15. The method of claim 8 , wherein the method of detection is a supervised classification algorithm such as decision trees, probabilistic boosting trees, support vector machines, or a combination thereof.
16. A high-throughput method for quantifying a feature in an image comprising the steps of:
a. selecting a domain swatch of data based upon a user specified domain knowledge of data within the image;
b. clustering the data within the image by conducting a frequency weighted mean shift of the data within the image to convergence to form a hierarchical plurality of layers of mean shifted data images, each having a different data resolution to form a hierarchical data pyramid;
c. segmenting the plurality of mean shifted data images to determine in each mean shifted data image within the hierarchical data pyramid data included within the swatch;
d. mapping the data included within the swatch spatially back to the image to create a final image; and
e. storing the final image on a storage medium for further analysis.
17. The method of claim 16 , wherein the image comprises an image of a tissue cylinder from within a tissue micro-array.
18. The method of claim 16 , wherein the image comprises an image of a tissue biopsy sample.
19. The method of claim 16 , wherein the domain swatch of data within the original image is selected from the group consisting of image color, intensity and texture.
20. The method of claim 16 , wherein said segmentation step comprises a normalized cuts algorithm.
21. The method of claim 16 , wherein the step of segmenting the plurality of mean shifted data images comprises segmenting the plurality of mean shifted data images to determine a color of interest in each mean shifted data image within the hierarchical data pyramid.