Bioimaging grid

• US 9,646,138 B2
• Filed: 12/31/2012
• Issued: 05/09/2017
• Est. Priority Date: 12/30/2011
• Status: Active Grant

• Current Assignee: Johns Hopkins University
• Original Assignee: Johns Hopkins University
• Inventors: Mori, Susumu, Miller, Michael I.
• Primary Examiner(s): TSAI, TSUNG YIN
• Application Number:
  US14/367,593
• Publication Number:
  US 20150302171A1
• Time in Patent Office: 1,590 Days
• Field of Search: None
• CPC Class Codes: G06T 2207/10072 Tomographic images
• CPC Class Codes: G06T 2207/20128 Atlas-based segmentation
• CPC Class Codes: G06T 2207/30016 Brain
• CPC Class Codes: G06T 7/0014 using an image reference approach
• CPC Class Codes: G16H 30/20 for handling medical images, e.g. DICOM, HL7 or PACS
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass

Abstract

A method may include standardizing an input image to produce a standardized input image; structurizing the standardized input image to be characterized by a plurality of anatomical properties; mapping the input image to one or more reference cases of a plurality of reference cases using the plurality of anatomical properties, wherein the one or more reference cases are associated with first non-image clinical information; and determining second non-image clinical information for the input image based on the one or more reference cases and the first non-image clinical information.

Claims

1. A computer-implemented method for evidence-based statistical diagnosis support, comprising:

receiving reference data for a plurality of reference cases represented in a feature space, wherein each reference case of said plurality of reference cases is based at least partially on corresponding imaging data and is associated with non-image clinical information;

segmenting the imaging data of each said reference case into a plurality of substructures for each of the plurality of reference cases;

determining an anatomical property associated with each substructure of the imaging data of the plurality of reference cases;

identifying a point in said feature space for each reference case based on at least said segmenting and said determining said anatomical property;

receiving data for a subject, said data comprising an input image;

segmenting the input image into a plurality of substructures;

determining an anatomical and non-anatomical property associated with each substructure of the plurality of substructures of the input image;

identifying a point in the feature space corresponding to the data for said subject based at least partially on the anatomical properties determined for said plurality of substructures of said standardized input image;

selecting a subset of reference cases from said plurality of reference cases based on distances of said point in said feature space corresponding to said data for said subject to each point in said feature space for each reference case; and

generating subject-specific clinical information based on the subset of reference cases using the corresponding non-image clinical information of each reference case in the subset of reference cases.

2. The method of claim 1, wherein the subset of reference cases further comprises multi-level clusters, and

wherein selecting said subset of reference cases from said plurality of reference cases is based on distances of said point in said feature space for said subject to at least one of the multi-level clusters.

3. The method of claim 2, wherein the selecting said subset of reference cases from said plurality of reference cases takes place only to one or more reference cases that represent one or more clusters to increase the efficiency of mapping.

4. The method of claim 1, wherein the subset of reference cases and the input image correspond to a same bodily segment.

5. The method of claim 1, wherein generating said subject-specific non-image clinical information based on the subset of reference cases and the corresponding non-image clinical information further comprises:

identifying non-image clinical information that corresponds to the subset of reference cases; and

selecting the subject-specific non-image clinical information from the identified non-image clinical information when an amount of the subject-specific non-image clinical information exceeds a threshold.

6. The method of claim 1, wherein the non-image clinical information comprises at least one of demographic information, one or more clinical symptoms, one or more functional test results, one or more diagnoses, and one or more prognoses.

7. The method of claim 1, further comprising:

presenting the subject-specific non-image clinical information as a report;

or

transmitting the subject-specific non-image clinical information to a recipient.

8. The method of claim 1, further comprising selecting one or more reference cases with higher level of anatomical similarity to increase selecting accuracy.

9. The method of claim 1, wherein the selecting is based on diffeomorphic image transformation.

10. The method of claim 1, further comprising:

standardizing the imaging data of the plurality of reference cases with respect to each other;

standardizing said input image of said subject with respect to said imaging data of the plurality of reference cases to provide a standardized input image,

wherein segmenting segments said standardized input image.

11. The method of claim 10, wherein the anatomical space is an N-dimension anatomical space.

12. The method of claim 11, wherein N is a number that corresponds to the number of segmented substructures.

13. The method of claim 1, wherein said feature space is an anatomical space.

14. A system for evidence-based statistical diagnosis support, comprising:

a memory; and

a processor connected to the memory and configured to;

receive reference data for a plurality of reference cases represented in an feature space, wherein each reference case of said plurality of reference cases is based at least partially on corresponding imaging data and is associated with non-image clinical information;

standardize the imaging data of the plurality of reference cases with respect to each other;

segment the imaging data of each said reference case into a plurality of substructures for each of the plurality of reference cases;

determine an anatomical property associated with each substructure of the imaging data of the plurality of reference cases;

identify a point in said feature space for each reference case based on at least said segmenting and said determining said anatomical property;

receive data for a subject, said data comprising an input image;

standardize said input image of said subject with respect to said imaging data of the plurality of reference cases to provide a standardized input image;

segment the standardized input image into a plurality of substructures;

determine an anatomical property associated with each substructure of the plurality of substructures;

identify a point in the feature space corresponding to the data for said subject based at least partially on the anatomical properties determined for said plurality of substructures of said standardized input image;

select a subset of reference cases from said plurality of reference cases based on distances of said point in said feature space corresponding to said data for said subject to each point in said feature space for each reference case; and

generate subject-specific clinical information based on the subset of reference cases using the corresponding non-image clinical information of each reference case in the subset of reference cases.

15. The system of claim 14, wherein the subset of reference cases further comprise multi-level clusters, and

wherein selecting said subset of reference cases from said plurality of reference cases is based on distances of said point in said feature space for said subject to at least one of the multi-level clusters.

16. The system of claim 14, wherein the subset of reference cases and the input image correspond to a same bodily segment.

17. The system of claim 14, wherein the processor is further configured to:

identify non-image clinical information that corresponds to the subset of reference cases; and

select the subject-specific non-image clinical information from the identified non-image clinical information when an amount of the subject-specific non-image clinical information exceeds a threshold.

18. The system of claim 14, wherein the non-image clinical information comprises at least one of demographic information, one or more clinical symptoms, one or more functional test results, one or more diagnoses, and one or more prognoses.

19. The system of claim 14, wherein the processor is further configured to:

present the subject-specific non-image clinical information as a report;

or

transmit the subject-specific non-image clinical information to a recipient.

20. A non-transitory computer readable storage medium for evidence-based statistical diagnosis support, the computer readable storage medium comprising instructions that if executed enable a computing system to:

receive reference data for a plurality of reference cases represented in a feature space, wherein each reference case of said plurality of reference cases is based at least partially on corresponding imaging data and is associated with non-image clinical information;

standardize the imaging data of the plurality of reference cases with respect to each other;

standardize said plurality of clinical images of said subjects with respect to said imaging data of the plurality of reference cases to provide a plurality of standardized clinical images;

segment the plurality of standardized clinical images into a plurality of substructures for each image of the plurality of standardized clinical images;

determine an anatomical property associated with each substructure of the imaging data of the plurality of reference cases;

identify a point in said feature space for each reference case based on at least said segmenting and said determining said anatomical property; and

store the plurality of reference cases in a database.

21. The computer readable storage medium of claim 20, further comprising instructions that if executed enable the computing system to:

receive data from a subject, said data comprising an input image;

standardize said input image of said subject with respect to said imaging data of the plurality of reference cases to provide a standardized input image;

segment the standardized input image into a plurality of substructures;

determine an anatomical property associated with each substructure of the plurality of substructures;

identify a point in the feature space corresponding to the data for said subject based at least partially on the anatomical properties determined for said plurality of substructures of said standardized input image;

select a subset of reference cases from said plurality of reference cases based on distances of said point in said feature space corresponding to said data for said subject to each point in said feature space for each reference case; and

generate subject-specific non-image clinical information based on the subset of reference cases using the corresponding non-image clinical information of each reference case in the subset of reference cases.

22. The computer readable storage medium of claim 21, wherein the one or more reference cases of the plurality of reference cases further comprise multi-level clusters, and

wherein mapping the input image to the one or more reference cases of the plurality of reference cases using the second plurality of anatomical properties further comprises instructions that if executed enable the computing system to:

map the input image to at least one of the multi-level clusters.

23. The computer readable storage medium of claim 21, wherein the subset of reference cases and the input image correspond to a same bodily segment.

24. The computer readable storage medium of claim 21, wherein generating said subject-specific non-image clinical information based on the subset of reference cases and the corresponding non-image clinical information further comprises instructions that if executed enable the computing system to:

identify non-image clinical information that corresponds to the subset of reference cases; and

select the subject-specific non-image clinical information from the identified non-image clinical information when an amount of the identified non-image clinical information exceeds a threshold.

25. The computer readable storage medium of claim 20, wherein said non-image clinical information comprises at least one of demographic information, one or more clinical symptoms, one or more functional test results, one or more diagnoses, and one or more prognoses.

26. The computer readable storage medium of claim 21, further comprising instructions that if executed enable the computing system to:

present the subject-specific non-image clinical information as a report;

or

transmit the subject-specific non-image clinical information to a recipient.

27. A computer-implemented method comprising:

mapping anatomical features of a patient of interest with respect to population data in a feature space from an image-based clinical database;

performing unsupervised analysis of the anatomical features of said patient using structured anatomical information matrices of the population data; and

performing supervised analysis of the anatomical features of said patient, wherein clinical information in the image-based clinical database is incorporated.

28. The method of claim 27, wherein an anatomy of the patient is mapped within a population based on anatomical similarity or similarity in clinically meaningful anatomical features using similarity indices.

29. A computer-implemented method comprising:

identifying a point in a feature space for each of image data stored in a clinical database;

determining distances among pairs of data in the features space from the image data stored in the database;

performing image structurization using image-vector or image-matrix conversion methods to produce anatomical image feature vectors or anatomical image feature matrices; and

based on the image feature vectors or image feature matrices, calculating a plurality of clusters based on similarity among the image feature vectors or image feature matrices.

30. The method of claim 29, wherein a representing case for each cluster is randomly selected within the cluster or calculated using population-averaging tools based on linear or non-linear image transformation.

31. A computer-implemented method to perform brain mapping and image parcellation using an iterative approach comprising:

identifying a point in a feature space for each of image data stored in a clinical database;

determining distances among pairs of data in the feature space from the image data stored in the database;

performing image structurization using image-vector or image-matrix conversion methods to produce anatomical image feature vectors or anatomical image feature matrices;

using the anatomical image feature vectors or anatomical image feature matrices to search similar references;

calculating a plurality of clusters based on similarity among the image feature vectors or image feature matrices;

if necessary, further searching inside proper clusters to obtain single or multiple references with similar anatomical features; and

iteratively searching by descending a cluster hierarchy to improve the mapping and parcellation accuracy.

32. The method of claim 31, wherein the iterative descending of a cluster hierarchy achieves improved accuracy for the population mapping by identifying similar anatomical features.