System and method of automatically detecting tissue abnormalities

Patent metadata

Field	Value
Publication number	US-9607392-B2
Application number	US-201214362354-A
Country	US
Kind code	B2
Filing date	Dec 5, 2012
Priority date	Dec 5, 2011
Publication date	Mar 28, 2017
Grant date	Mar 28, 2017

Abstract

A method of automatically detecting tissue abnormalities in images of a region of interest of a subject includes obtaining first image data for the region of interest of the subject, normalizing the first image data based on statistical parameters derived from at least a portion of the first image data to provide first normalized image data, obtaining second image data for the region of interest of the subject, normalizing the second image data based on statistical parameters derived from at least a portion of the second image data to provide second normalized image data, processing the first and second normalized image data to provide resultant image data, and generating a probability map for the region of interest based on the resultant image data and a predefined statistical model. The probability map indicates the probability of at least a portion of an abnormality being present at locations within the region of interest.

First claim

We claim:

1. A method of automatically detecting tissue abnormalities in images of a region of interest of a subject, comprising:
obtaining first image data for said region of interest of said subject;
segmenting said first image data into a plurality of sub-images corresponding to a plurality of anatomical structures, each sub-image comprising a plurality of image elements;
normalizing each sub-image of said first image data based on statistical parameters derived from said plurality of image elements within each corresponding sub-image to provide first normalized image data;
obtaining second image data for said region of interest of said subject;
segmenting said second image data into a plurality of sub-images corresponding to a plurality of anatomical structures, each sub-image comprising a plurality of image elements;
normalizing each sub-image of said second image data based on statistical parameters derived from said plurality of image elements within each corresponding sub-image to provide second normalized image data;
processing said first and second normalized image data to provide resultant image data; and
generating a probability map for said region of interest based on said resultant image data and a predefined statistical model,
wherein said probability map indicates the probability of at least a portion of an abnormality being present at locations within said region of interest.

2. The method of claim 1 , wherein said predefined statistical model is a logistic regression model.

3. The method of claim 1 , wherein said first image data and said second image data correspond to images taken at different times.

4. The method of claim 1 , wherein said first image data and said second image data are for a single type of imaging modality.

5. The method of claim 4 , wherein said single type of imaging modality is one of an MRI, X-ray computed tomography, positron emission tomography, single-photon emission computed tomography, or ultrasound.

6. The method of claim 4 , wherein said single type of imaging modality is one of a proton density, fluid-attenuated inversion recovery, T 2 -weighted, or T 1 -weighted MRI imaging modality.

7. The method of claim 1 , further comprising:
obtaining third image data for said region of interest of said subject;
normalizing said third image data based on statistical parameters derived from at least a portion of said third image data to provide third normalized image data;
obtaining fourth image data for said region of interest of said subject;
normalizing said fourth image data based on statistical parameters derived from at least a portion of said fourth image data to provide fourth normalized image data;
processing said third and fourth normalized image data to provide a second resultant image data,
wherein said generating said probability map for said region of interest is further based on said second resultant image data.

8. The method of claim 7 , wherein said third image data and said fourth image data are for a single type of imaging modality that is a different imaging modality than said single imaging modality of said first and second imaging data.

9. A non-transitory computer-readable medium comprising machine-executable code for automatically detecting tissue abnormalities in images of a region of interest of a subject, said machine-executable code, when executed by a computer, causes the computer to:
obtain first image data for said region of interest of said subject;
segment said first image data into a plurality of sub-images corresponding to a plurality of anatomical structures, each sub-image comprising a plurality of image elements;
normalize each sub-image of said first image data based on statistical parameters derived from said plurality of image elements within each corresponding sub-image to provide first normalized image data;
obtain second image data for said region of interest of said subject;
segment said second image data into a plurality of sub-images corresponding to a plurality of anatomical structures, each sub-image comprising a plurality of image elements;
normalize each sub-image of said second image data based on statistical parameters derived from said plurality of image elements within each corresponding sub-image to provide second normalized image data;
process said first and second normalized image data to provide resultant image data; and
generate a probability map for said region of interest based on said resultant image data and a predefined statistical model,
wherein said probability map indicates the probability of at least a portion of an abnormality being present at locations within said region of interest.

10. The non-transitory computer-readable medium of claim 9 , wherein said predefined statistical model is a logistic regression model.

11. The non-transitory computer-readable medium of claim 9 , wherein said first image data and said second image data correspond to images taken at different times.

12. The non-transitory computer-readable medium of claim 9 , wherein said first image data and said second image data are for a single type of imaging modality.

13. The non-transitory computer-readable medium of claim 12 , wherein said single type of imaging modality is one of an MRI, X-ray computed tomography, positron emission tomography, single-photon emission computed tomography, or ultrasound.

14. The non-transitory computer-readable medium of claim 12 , wherein said single type of imaging modality is one of a proton density, fluid-attenuated inversion recovery, T 2 -weighted, or T 1 -weighted MRI imaging modality.

15. The non-transitory computer-readable medium of claim 9 , wherein said machine-executable code, when executed by said computer, further causes the computer to:
obtain third image data for said region of interest of said subject;
normalize said third image data based on statistical parameters derived from at least a portion of said third image data to provide third normalized image data;
obtain fourth image data for said region of interest of said subject;
normalize said fourth image data based on statistical parameters derived from at least a portion of said fourth image data to provide fourth normalized image data; and
process said third and fourth normalized image data to provide a second resultant image data,
wherein said generating said probability map for said region of interest is further based on said second resultant image data.

16. The non-transitory computer-readable medium of claim 15 , wherein said third image data and said fourth image data are for a single type of imaging modality that is a different imaging modality than said single imaging modality of said first and second imaging data.

17. A system for automatically detecting tissue abnormalities in images of a region of interest of a subject, comprising:
a signal processor configured to:
obtain first image data for said region of interest of said subject;
segment said first image data into a plurality of sub-images corresponding to a plurality of anatomical structures, each sub-image comprising a plurality of image elements;
normalize each sub-image of said first image data based on statistical parameters derived from said plurality of image elements within each corresponding sub-image to provide first normalized image data;
obtain second image data for said region of interest of said subject;
segment said second image data into a plurality of sub-images corresponding to a plurality of anatomical structures, each sub-image comprising a plurality of image elements;

Assignees

• Univ Johns Hopkins
• Us Health

Inventors

• Crainiceanu Ciprian M
• Sweeney Elizabeth M
• Shinohara Russell T
• Goldsmith Arthur J
• Reich Daniel
• Shea Colin

Classifications

• G06T7/0012 Primary

  Biomedical image inspection · CPC title

• A61B8/00

  Diagnosis using ultrasonic, sonic or infrasonic waves · CPC title

• G06T2207/10088

  Magnetic resonance imaging [MRI] · CPC title

• G06T2207/20224

  Image subtraction · CPC title

• G01R33/4806

  Functional imaging of brain activation · CPC title