Method of analyzing a medical image

• US 8,965,089 B2
• Filed: 05/07/2014
• Issued: 02/24/2015
• Est. Priority Date: 02/23/2010
• Status: Active Grant

• Current Assignee: LOMA Linda University, Regents of the University of California (University of California)
• Original Assignee: LOMA Linda University, Regents of the University of California (University of California)
• Inventors: Ghosh, Nirmalya, Ashwal, Stephen, Obenaus, Andre, Bhanu, Bir
• Primary Examiner(s): Seth, Manav
• Application Number:
  US14/272,299
• Publication Number:
  US 20140241608A1
• Time in Patent Office: 293 Days
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/10088 Magnetic resonance imaging [MRI]
• CPC Class Codes: G06T 2207/10104 Positron emission tomography [PET]
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20036 Morphological image processing
• CPC Class Codes: G06T 2207/20072 Graph-based image processing
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/30016 Brain
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 2207/30052 Implant; Prosthesis
• CPC Class Codes: G06T 2207/30056 Liver; Hepatic
• CPC Class Codes: G06T 2207/30061 Lung
• CPC Class Codes: G06T 2207/30084 Kidney; Renal
• CPC Class Codes: G06T 2207/30088 Skin; Dermal
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06T 7/155 involving morphological operators
• CPC Class Codes: G06T 7/162 : involving graph-based methods
• CPC Class Codes: G06V 10/25 : Determination of region of interest [ROI] or a volume of interest [VOI]
• CPC Class Codes: G06V 10/28 : Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns

Abstract

A method of analyzing a medical image, where the medical image comprises one or more than one region of interest, and where the method comprises a) providing the medical image comprising a set of actual image values; b) rescaling the actual image values to produce corresponding rescaled image values and to produce a rescaled image from the rescaled image values; c) deriving a histogram of the rescaled image values; d) using the histogram to derive an adaptive segmentation threshold; e) using the adaptive segmentation threshold to recursively split the rescaled image; f) terminating the recursive splitting of the sub(sub) images using one or more than one predetermined criteria; and g) identifying one sub(sub) image in the terminated Hierarchical Region Splitting Tree which comprises the region of interest.

Claims

1. A method of analyzing a medical image, the medical image comprising one or more than one region of interest, the method comprising:

a) providing the medical image comprising a set of actual image values;

b) rescaling the actual image values to produce corresponding rescaled image values and to produce a rescaled image from the rescaled image values;

c) deriving a histogram of the rescaled image values;

d) using the histogram to derive an adaptive segmentation threshold that can be used to split the rescaled image into two sub-images, a first sub-image with intensities at or below the adaptive segmentation threshold and a second sub-image with intensities above the adaptive segmentation threshold, or a first sub-image with intensities below the adaptive segmentation threshold and a second sub-image with intensities at or above the adaptive segmentation threshold, or a first sub-image with intensities below the adaptive segmentation threshold and a second sub-image with intensities above the adaptive segmentation threshold;

e) using the adaptive segmentation threshold to recursively split the rescaled image to generate a Hierarchical Region Splitting Tree of sub(sub) images based on consistency of the rescaled image values of the rescaled image;

f) terminating the recursive splitting of the sub(sub) images using one or more than one predetermined criteria thereby completing the Hierarchical Region Splitting Tree; and

g) identifying one sub(sub) image in the terminated Hierarchical Region Splitting Tree which comprises the region of interest;

the method further comprising performing a secondary rescaling of the rescaled image values of every rescaled sub(sub) image in the Hierarchical Region Splitting Tree back to the actual image values present in the medical image to create a secondary rescaled medical image, thereby producing a secondarily rescaled sub(sub) image comprising the region of interest;

where the rescaled image values fit in [0,255] unsigned 8-bit integer range;

where the predetermined criteria is selected from the group consisting of area threshold=50 pixels and (standard deviation threshold=10 rscVals (StdDevTh=10 rscVals) and kurtosis threshold=1.5);

where the region of interest is a representation of an abnormality in the living human tissue, and where the method further comprises quantifying the abnormality in the living human tissue;

where the method further comprises performing a secondary rescaling of the rescaled image values in every rescaled sub(sub) image in the Hierarchical Region Splitting Tree back to the actual image values present in the medical image to create a secondary rescaled medical image, and determining an image value or a set of image values of actual image values in the medical image after the secondary rescaling, where the image value or a set of image values of actual image values determined identifies the abnormality represented in the medical image for the modality being used to generate the medical image provided; and

where the method further comprises preparing a mask of the sub(sub) image containing the representation of the abnormality, and cleaning the mask to remove small outlier regions to generate a cleaned mask of the sub(sub) image containing the representation of the abnormality.

2. The method of claim 1, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is two-dimensional.

3. The method of claim 1, where the secondarily rescaled sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is three-dimensional.

4. The method of claim 1, where the secondarily rescaled one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is two-dimensional.

5. The method of claim 1, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is three-dimensional.

6. The method of claim 1, where the medical image is selected from the group consisting of a computed tomography scan, a magnetic resonance image, a positron emission tomography scan and an X-ray.

7. The method of claim 1, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is two-dimensional.

8. The method of claim 1, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is three-dimensional.

9. The method of claim 1, where the region of interest is a representation of an injury to living human tissue and the method detects the representation of the injury.

10. The method of claim 9, where the method qualifies the representation of the injury.

11. The method of claim 1, where the medical image is of human tissue selected from the group consisting of brain, heart, intestines, joints, kidneys, liver, lungs and spleen.

12. The method of claim 1, where the medical image provided is in a hard copy form, and where the method further comprises preparing a digital form of the medical image before providing the medical image.

13. The method of claim 1, where the abnormality is selected from the group consisting of a genetic malformation and an injury.

14. The method of claim 1, where determining the image value or a set of image values of actual image values is made before the step of providing the medical image.

15. The method of claim 1, where determining the image value or a set of image values of actual image values is made after the step of providing the medical image.

16. The method of claim 1, where the medical image is a magnetic resonance image and the modality being used to generate the medical image provided is selected from the group consisting of an apparent diffusion coefficient map, a magnetic susceptibility map and a T2 map.

17. A method of detecting a core of an injury and detecting a penumbra of an injury in living human tissue, and distinguishing the core from the penumbra, the method comprising:

a) detecting one sub(sub) image (the injury sub-image) in the terminated Hierarchical Region Splitting Tree comprising the region of interest according to claim 1, where the region of interest represents the injury;

b) determining the mask of the injury;

c) determining a sub-tree below the detected injury sub-image using the injury sub-image as the root of the sub-tree;

d) determining soft threshold image values for separating the core from the penumbra;

e) comparing the soft threshold image values inside the sub-tree to find the penumbra and a mask of the penumbra; and

f) determining the mask of the core by subtracting the mask of the penumbra from the mask of the injury.

18. The method of claim 17, further comprising determining different gradations of the core and the penumbra.

19. The method of claim 1, further comprising quantifying the spatiotemporal evolution of an injury in living human tissue.

20. A method of detecting the effects of endogenous or implanted stem cells on living human tissue, the method comprising:

a) determining magnetic resonance image values of labeled and implanted stem cells;

b) detecting the stem cells outside of the region of interest using a method according to claim 1; and

c) detecting the stem cells inside of the region of interest using a method according to claim 1.

21. The method of claim 19, further comprising quantifying spatiotemporal activities of implanted labeled stem cells in the living human tissue.

22. A method of analyzing a medical image, the medical image comprising one or more than one region of interest, the method comprising:

a) configuring at least one processor to perform the functions of;

1) providing the medical image comprising a set of actual image values;

2) rescaling the actual image values to produce corresponding rescaled image values and to produce a rescaled image from the rescaled image values;

3) deriving a histogram of the rescaled image values;

4) using the histogram to derive an adaptive segmentation threshold that can be used to split the rescaled image into two sub-images, a first sub-image with intensities at or below the adaptive segmentation threshold and a second sub-image with intensities above the adaptive segmentation threshold, or a first sub-image with intensities below the adaptive segmentation threshold and a second sub-image with intensities at or above the adaptive segmentation threshold, or a first sub-image with intensities below the adaptive segmentation threshold and a second sub-image with intensities above the adaptive segmentation threshold;

5) using the adaptive segmentation threshold to recursively split the rescaled image to generate a Hierarchical Region Splitting Tree of sub(sub) images based on consistency of the rescaled image values of the rescaled image;

6) terminating the recursive splitting of the sub(sub) images using one or more than one predetermined criteria thereby completing the Hierarchical Region Splitting Tree; and

7) identifying one sub(sub) image in the terminated Hierarchical Region Splitting Tree which comprises the region of interest;

the method further comprising performing a secondary rescaling of the rescaled image values of every rescaled sub(sub) image in the Hierarchical Region Splitting Tree back to the actual image values present in the medical image to create a secondary rescaled medical image, thereby producing a secondarily rescaled sub(sub) image comprising the region of interest;

where the rescaled image values fit in [0,255] unsigned 8-bit integer range;

where the predetermined criteria is selected from the group consisting of area threshold=50 pixels and (standard deviation threshold=10 rscVals (StdDevTh=10 rscVals) and kurtosis threshold=1.5);

where the region of interest is a representation of an abnormality in the living human tissue, and where the method further comprises quantifying the abnormality in the living human tissue;

where the method further comprises performing a secondary rescaling of the rescaled image values in every rescaled sub(sub) image in the Hierarchical Region Splitting Tree back to the actual image values present in the medical image to create a secondary rescaled medical image, and determining an image value or a set of image values of actual image values in the medical image after the secondary rescaling, where the image value or a set of image values of actual image values determined identifies the abnormality represented in the medical image for the modality being used to generate the medical image provided;

where the medical image is a magnetic resonance image and the modality being used to generate the medical image provided is selected from the group consisting of an apparent diffusion coefficient map, a magnetic susceptibility map and a T2 map; and

where the method further comprises preparing a mask of the sub(sub) image containing the representation of the abnormality, and cleaning the mask to remove small outlier regions to generate a cleaned mask of the sub(sub) image containing the representation of the abnormality.

23. The method of claim 22, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is two-dimensional.

24. The method of claim 22, where the secondarily rescaled sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is three-dimensional.

25. The method of claim 22, where the secondarily rescaled one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is two-dimensional.

26. The method of claim 22, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is three-dimensional.

27. The method of claim 22, where the medical image is selected from the group consisting of a computed tomography scan, a magnetic resonance image, a positron emission tomography scan and an X-ray.

28. The method of claim 22, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is two-dimensional.

29. The method of claim 22, where the one sub(sub) image in the terminated Hierarchical Region Splitting Tree comprising the region of interest is three-dimensional.

30. The method of claim 22, where the region of interest is a representation of an injury to living human tissue and the method detects the representation of the injury.

31. The method of claim 30, where the method qualifies the representation of the injury.

32. The method of claim 22, where the medical image is of human tissue is selected from the group consisting of brain, heart, intestines, joints, kidneys, liver, lungs and spleen.

33. The method of claim 22, where the medical image provided is in a hard copy form, and where the method further comprises preparing a digital form of the medical image before providing the medical image.

34. The method of claim 33, where the abnormality is selected from the group consisting of a genetic malformation and an injury.

35. The method of claim 33, where determining the image value or a set of image values of actual image values is made before the step of providing the medical image.

36. The method of claim 33, where determining the image value or a set of image values of actual image values is made after the step of providing the medical image.

37. The method of claim 33, where the medical image is a magnetic resonance image and the modality being used to generate the medical image provided is selected from the group consisting of an apparent diffusion coefficient map, a magnetic susceptibility map and a T2 map.