NETWORK CYCLE FEATURES IN RELATIVE NEIGHBORHOOD GRAPHS

• US 20130243292A1
• Filed: 02/28/2013
• Published: 09/19/2013
• Est. Priority Date: 03/01/2012
• Status: Active Grant

• Current Assignee: BETH Israel Deaconess Medical Center (Caregroup Incorporated), Siemens Corp. (Siemens AG)
• Original Assignee: Siemens Corp. (Siemens AG)
• Inventors: Khurd, Parmeshwar, Kamen, Ali, Sundar, Hari, Frangioni, John V., Gibbs-Strauss, Summer
• Granted Patent: US 9,047,660 B2
• Time in Patent Office: Days
• US Class Current: 382/131
• CPC Class Codes: A61B 6/5229 combining image data of a patient, e.g. combining a functional image with an anatomical image
• CPC Class Codes: G06T 17/20 Finite element generation, e.g. wire-frame surface description, tesselation
• CPC Class Codes: G06T 2207/10056 Microscopic image
• CPC Class Codes: G06T 2207/20072 Graph-based image processing
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/344 involving models
• CPC Class Codes: G06V 10/426 Graphical representations
• CPC Class Codes: G06V 20/695 Preprocessing, e.g. image segmentation

Abstract

Methods for analyzing biomedical data include: (a) obtaining macroscopic imaging data; (b) obtaining histopathological imaging data; (c) executing a parallel algorithm stored on a non-transient computer-readable medium to compute one or a plurality of network cycle features of a relative neighborhood graph derived from the histopathological imaging data; (d) registering the macroscopic imaging data and the histopathological imaging data; and (e) correlating the macroscopic imaging data and the network cycle features. Systems for analyzing biomedical data and computer readable storage media are described.

Claims

1. A method for analyzing biomedical data comprising;

obtaining macroscopic imaging data;

obtaining histopathological imaging data;

executing a parallel algorithm stored on a non-transient computer-readable medium to compute one or a plurality of network cycle features of a relative neighborhood graph derived from the histopathological imaging data;

registering the macroscopic imaging data and the histopathological imaging data; and

correlating the macroscopic imaging data and the network cycle features.

2. The method of claim 1 wherein the network cycle features are selected from the group consisting of cycle order, cycle perimeter, and a combination thereof.

3. The method of claim 1 further comprising generating the relative neighborhood graph from the histopathological imaging data by parallel computation.

4. The method of claim 3 further comprising:

quadtree-based partitioning of nuclei coordinates obtained from the histopathological imaging data across a plurality of processes;

Morton ordering of the nuclei coordinates to optimize load balance;

communicating a shared boundary layer;

triangulating by each process the shared boundary layer using Delaunay triangulations;

adding remaining local points to the respective triangulations of each process;

pruning the triangulations using an Urquhart approximation to generate the relative neighborhood graphs;

partitioning duplicated triangles in the shared boundary layer;

marking a longest edge of each the triangles;

merging polygons sharing marked edges locally; and

merging boundary polygons.

5. The method of claim 1 wherein the biomedical data are indicative of whether a disease is present and/or a severity of the disease.

6. The method of claim 5 wherein the disease comprises cancer.

7. The method of claim 5 wherein the disease comprises prostate cancer.

8. The method of claim 1 wherein the macroscopic imaging data are selected from the group consisting of magnetic resonance data, computed tomography data, positron emission tomography data, single photon emission tomography data, and combinations thereof.

9. The method of claim 1 wherein the macroscopic imaging data comprise dynamic contrast enhanced magnetic resonance data.

10. A system for analyzing biomedical data comprising:

a memory operable to store macroscopic imaging data and histopathological imaging data;

a processor operable to (a) generate a relative neighborhood graph from the histopathological imaging data by parallel computation;

(b) compute one or a plurality of network cycle features of the relative neighborhood graph by parallel computation; and

(c) correlate the macroscopic imaging data and the network cycle features; and

a display operable to display an analysis result derived from correlation of the macroscopic imaging data and the network cycle features.

11. The system of claim 10 wherein the network cycle features are selected from the group consisting of cycle order, cycle perimeter, and a combination thereof.

12. The system of claim 10 further comprising a user input operable to receive input from a system user.

13. The system of claim 10 wherein the biomedical data are indicative of whether a disease is present and/or a severity of the disease.

14. The system of claim 13 wherein the disease comprises cancer.

15. The system of claim 13 wherein the disease comprises prostate cancer.

16. The system of claim 10 wherein the macroscopic imaging data are selected from the group consisting of magnetic resonance data, computed tomography data, positron emission tomography data, single photon emission tomography data, and combinations thereof.

17. The system of claim 10 wherein the macroscopic imaging data comprise dynamic contrast enhanced magnetic resonance data.

18. In a non-transitory computer readable storage medium having stored therein data representing instructions executable by a programmed processor for biomedical study, the storage medium comprising instructions for:

generating a relative neighborhood graph from histopathological imaging data by parallel computation;

computing one or a plurality of network cycle features of the relative neighborhood graph by parallel computation; and

correlating the macroscopic imaging data and the network cycle features.

19. The non-transitory computer readable storage medium of claim 18 wherein the storage medium further comprises instructions for:

quadtree-based partitioning of nuclei coordinates obtained from the histopathological imaging data across a plurality of processes;

Morton ordering of the nuclei coordinates to optimize load balance;

creation of a shared boundary layer, wherein each of the processes Delaunay triangulates the shared boundary layer, and wherein remaining local points are added to each process'"'"'s respective triangulations; and

pruning the triangulations to generate the relative neighborhood graph.

20. The non-transitory computer readable storage medium or claim 18 wherein the storage medium further comprises instructions for:

preparing an Urquhart graph from the Delaunay triangulations;

partitioning duplicated triangles in the shared boundary layer;

marking a longest edge of each triangle;

merging polygons sharing marked edges locally; and

merging boundary polygons.

21. A method for analyzing biomedical data comprising:

(a) obtaining histopathological imaging data;

(b) executing a parallel algorithm stored on a non-transient computer-readable medium to compute one or a plurality of network cycle features of a relative neighborhood graph derived from the histopathological imaging data; and

(c) predicting a presence and/or severity of a disease using the network cycle features, and/or correlating graph-theoretic microscopic biomarkers with the presence and/or severity of the disease.

22. The method of claim 21 wherein the disease comprises cancer.