Atlas-based analysis for image-based anatomic and functional data of organism

• US 8,838,201 B2
• Filed: 06/22/2011
• Issued: 09/16/2014
• Est. Priority Date: 06/22/2010
• Status: Active Grant

• Current Assignee: Johns Hopkins University
• Original Assignee: Johns Hopkins University
• Inventors: Mori, Susumu, Faria, Andreia V., Miller, Michael I., Oishi, Kenichi
• Primary Examiner(s): MEHTA, PARIKHA SOLANKI
• Application Number:
  US13/703,859
• Publication Number:
  US 20130102877A1
• Time in Patent Office: 1,182 Days
• Field of Search: 600/410, 600/411, 600/416, 382/128, 382/173, 382/190, 382/209, 382/224, 703/11
• US Class Current: 600/410
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: A61B 6/03 Computed tomography [CT] echo-tomography A61B8/14
• CPC Class Codes: A61B 6/5217 extracting a diagnostic or physiological parameter from medical diagnostic data
• CPC Class Codes: G01R 33/5608 Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels image data processing or generation, in general G06T
• 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 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/149 involving deformable models, e.g. active contour models
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment

Abstract

A non-invasive imaging system, including an imaging scanner suitable to generate an imaging signal from a tissue region of a subject under observation, the tissue region having at least one anatomical substructure and more than one constituent tissue type; a signal processing system in communication with the imaging scanner to receive the imaging signal from the imaging scanner; and a data storage unit in communication with the signal processing system, wherein the data storage unit is configured to store a parcellation atlas comprising spatial information of the at least one substructure in the tissue region, wherein the signal processing system is adapted to: reconstruct an image of the tissue region based on the imaging signal; parcellate, based on the parcellation atlas, the at least one anatomical substructure in the image; segment the more than one constituent tissue types in the image; and automatically identify, in the image, a portion of the at least one anatomical substructure that correspond to one of the more than one constituent tissue type.

Claims

1. A non-invasive imaging system, comprising:

an imaging scanner configured to generate an imaging signal from a tissue region of a subject under observation, the tissue region having at least one anatomical substructure and more than one constituent tissue type;

a signal processing system in communication with said imaging scanner configured to receive the imaging signal from said imaging scanner; and

a data storage unit in communication with said signal processing system,

wherein said data storage unit is configured to store a parcellation atlas comprising spatial information of said at least one anatomical substructure in the tissue region,

wherein said signal processing system is adapted to;

reconstruct an image of said tissue region based on said imaging signal;

parcellate, based on the parcellation atlas, the at least one anatomical substructure in said image;

segment the more than one constituent tissue type in said image to provide a plurality of segmented constituent tissue types, wherein the parcellation atlas includes at least one of imaging data (i) of the more than one constituent tissue type, (ii) of at least one of a plurality of imaging modalities and a plurality of imaging contrasts corresponding to one subject, and (iii) corresponding to a plurality of subjects;

automatically identify, in said image, a portion of the at least one anatomical substructure that corresponds to one of the more than one constituent tissue type; and

average all image voxels in the identified constituent tissue type to increase a signal-to-noise ratio of the automatically identified portion,

wherein averaging the image voxels at least one of increases image resolution of said image and decreases scanning time from said imaging scanner, and

wherein the at least one anatomical substructure is defined in one of the image modalities or imaging contrasts which has a higher signal-to-noise or resolution and applied to other image modalities or imaging contrasts.

2. The non-invasive imaging system according to claim 1, wherein said imaging scanner is a magnetic resonance imaging (MRI) scanner, a computed tomography (CT) imaging scanner, a positron emission tomography (PET) imaging scanner, a single positron emission computed tomography (SPECT) imaging scanner, or a combination thereof.

3. The non-invasive imaging system according to claim 1, wherein said imaging signal is capable of providing at least one contrast mechanism that delineates one of said more than one constituent tissue type of said tissue region.

4. The imaging system according to claim 1, wherein said signal processing system is further adapted to automatically diagnose whether a tissue abnormality is present in said tissue region of said subject by analyzing the portion of the at least one anatomical substructure that correspond to one of the more than one constituent tissue types.

5. The imaging system according to claim 1, further comprising:

a viewing station or a console station configured to display the portion of the at least one anatomical substructure by superimposing said portion on said at least one anatomical substructure.

6. A workstation, comprising:

a receiving engine adapted to receive an input image representing a tissue region of a subject, and

a parcellation atlas comprising spatial information of at least one anatomical substructure in the tissue region;

a normalizing engine configured to provide a parcellated image by registering the parcellation atlas, via a transformation, to the input image; and

a computing engine configured to;

automatically identify a portion of the at least one anatomical substructure, said portion corresponding to one constituent tissue type of said tissue region, wherein the parcellated atlas includes at least one of imaging data (i) of the more than one constituent tissue type, (ii) of at least one of a plurality of imaging modalities and a plurality of imaging contrasts corresponding to one subject, and (iii) corresponding to a plurality of subjects, and

average all image voxels in the identified constituent tissue type to increase a signal-to-noise ratio of the automatically identified portion,

wherein the at least one anatomical substructure is defined in one of the modalities or imaging contrasts and applied to other imaging modalities or imaging contrasts, and

wherein averaging the image voxels at least one of increases image resolution of said image and decreases scanning time from said imaging scanner.

7. A workstation of claim 6, further comprising an analyzing engine configured to analyze said portion by combining image voxels corresponding to said portion.

8. The workstation according to claim 6, further comprising:

a visualizing engine adapted to display the parcellated image with said portion identified.

9. A method of analyzing an input image, comprising:

receiving, from one of an imaging system, a workstation, or a data storage device, an input image representing a tissue region of a subject,

wherein the input image comprises a plurality of image voxels,

wherein said tissue region has a plurality of anatomical substructures and a plurality of constituent tissue types, and

wherein at least one of the plurality of anatomical substructures comprises at least two constituent tissue types from the plurality of constituent tissue types;

providing a parcellation atlas of said tissue region comprising spatial information of said anatomical substructures, wherein said parcellation atlas is from one of the first data storage device, or a second data storage device;

generating a parcellated image by registering said parcellation atlas to the input image via a transformation, wherein the anatomical substructures of said tissue region are parcellated in the parcellated image;

providing a segmented parcellated image by identifying image voxels of one of the anatomical substructures in the parcellated image that correspond to one of said plurality of constituent tissue types to provide a plurality of segmented constituent tissue types, wherein the parcellation atlas includes at least one of imaging data (i) of the more than one segmented constituent tissue types, (ii) of at least one of a plurality of imaging modalities and a plurality of imaging contrasts corresponding to one subject to generate the parcellation atlas, and (iii) corresponding to a plurality of subjects; and

averaging all image voxels in the identified constituent tissue type to increase a signal-to-noise ratio of the segmented parcellated image,

wherein averaging the image voxels at least one of increases image resolution of said image and decreases scanning time from said imaging scanner, and

wherein the at least one anatomical substructure is defined in one of the image modalities or imaging contrasts which has a higher signal-to-noise or resolution and applied to other imaging modalities or imaging contrasts.

10. The method of claim 9, further comprising:

using the parcellated image or the segemented parcellated segmented image to improve a signal-to-noise ratio or an accuracy associated with an analysis of said input image.

11. The method of claim 9, further comprising:

analyzing said segmented parcellated image by grouping the image voxels of one of the anatomical substructures in the parcellated image that correspond to one of said plurality of constituent tissue types.

12. The method of claim 11, further comprising:

computing a quantity of the image voxels that correspond to one of said plurality of constituent tissue type and one of said anatomical substructures.

13. The method of claim 12, wherein said quantity is one of a mean, a sum, or a median.

14. The method of claim 12, further comprising:

automatically diagnosing a tissue abnormality of said tissue region based on said quantity.

15. The method of claim 14, wherein said tissue abnormality is one of a tissue atrophy, a stroke infarction, or a tumor.

16. The method of claim 14, wherein said tissue abnormality corresponds to an abnormality in one of a Magnetic Resonance Imaging (MRI) relaxation time change, a diffusion MRI parameter change, a Positron Emission Tomography (PET) image intensity, a Computed Tomography (CT) image intensity, or a Magnetic Resonance Spectroscopy (MRS) metabolite concentration.

17. The method of claim 11, further comprising:

increasing a quantitative index associated with said analyzing.

18. The method of claim 17, wherein said quantitative index is one of a sensitivity, a specificity, or a reproducibility.

19. The method of claim 9, wherein said identifying further comprises:

providing a segmentation atlas of said tissue region, said segmentation atlas comprising spatial information of said plurality of constituent tissue types; and

segmenting the input image by registering said segmentation atlas of said tissue region, via a transformation, to said input image.

20. The method of claim 9, wherein said input image is one of a Magnetic Resonance Imaging (MRI) image, a functional MRI (fMRI) image, a Magnetic Resonance Spectroscopy (MRS) image, a Positron Emission Tomography (PET) image, or a Computed Tomography (CT) image.

21. The method of claim 9, wherein the parcellation atlas has a corresponding spatial resolution that is higher than the input image'"'"'s spatial resolution.

22. The method of claim 21, further comprising:

compensating a partial volume effect in said input image by using the parcellation atlas to generate the parcellated image.

23. The method of claim 22, wherein said using the parcellation atlas further comprises:

estimating a weight of one of said plurality of anatomical substructures in each image voxel in the input image; and

applying said weight to said parcellated image such that the parcellated image includes the weight.

24. The method of claim 9, further comprising:

using the segmented parcellated image to measure time-dependent image intensity changes in a functional MR1 (fMRI) study.

25. The method of claim 9, further comprising using the segmented parcellated image for a correlation analysis of a resting-state fMRI study.

26. The method of claim 9, wherein said transformation is a Large Deformation Diffeomorphic Metric Mapping (LDDMM).

27. A non-invasive imaging system, comprising:

an imaging scanner configured to generate an imaging signal from a tissue region of a subject under observation, the tissue region having at least one anatomical substructure and more than one constituent tissue type;

a signal processing system in communication with said imaging scanner configured to receive the imaging signal from said imaging scanner; and

a data storage unit in communication with said signal processing system,

wherein said data storage unit is configured to store a parcellation atlas comprising spatial information of said at least one anatomical substructure in the tissue region,

wherein said signal processing system is adapted to;

reconstruct an image of said tissue region based on said imaging signal;

parcellate, based on the parcellation atlas, the at least one anatomical substructure in said image to generate a parcellated image;

segment the more than one constituent tissue type in said image to generate a segmented image;

merge the parcellated image and the segmented image to generate a parcellated, segmented image; and

automatically identify, in said parcellated, segmented image, a portion of the at least one anatomical substructure that corresponds to one of the more than one constituent tissue type.