Non-invasive systems and methods for the determination of cardiac injury using a characterizing portion of a voxel histogram

• US 7,747,308 B2
• Filed: 02/02/2006
• Issued: 06/29/2010
• Est. Priority Date: 02/06/2004
• Status: Active Grant

• Current Assignee: Wake Forest University Health Sciences
• Original Assignee: Wake Forest University Health Sciences
• Inventors: Lane, Kimberly, Morgan, Tim, D'Agostino, Ralph B., Hamilton, Craig A., Hundley, William Gregory, Torti, Frank
• Primary Examiner(s): AZARIAN, SEYED H
• Application Number:
  US11/346,527
• Publication Number:
  US 20080004521A1
• Time in Patent Office: 1,608 Days
• Field of Search: 382/100, 382/128, 382/129, 382/130, 382/131, 382/132, 382/133, 382/134, 382/154, 382/162, 382/168, 382/181, 382/196, 382/203, 382/219, 382/232, 382/255, 382/274, 382/276, 382/286, 382/305, 382/320, 435/7.92, 536/23.2, 600/529, 600/407, 600/410, 600/420, 600/414
• US Class Current: 600/414
• CPC Class Codes: A61B 5/055 involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
• CPC Class Codes: A61B 6/506 for diagnosis of nerves
• CPC Class Codes: G06T 2207/10088 Magnetic resonance imaging [MRI]
• CPC Class Codes: G06T 2207/10096 Dynamic contrast-enhanced magnetic resonance imaging [DCE-MRI]
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06V 10/758 Involving statistics of pixels or of feature values, e.g. histogram matching
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

Methods, systems and circuits predict cardiotoxicity induced cardiac injury prior to an irreversible state by electronically generating at least one histogram of mean intensities of voxels/pixels in an MRI image of a left ventricle myocardium and electronically determining a likelihood of cardiac injury due to cardiotoxicity based on data from the at least one histogram.

Claims

1. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

electronically determining, for a plurality of regions of interest in a medical image of a heart, respective locations of voxels in three-dimensional space and an associated respective measure of intensity for each of the voxels;

electronically identifying whether high intensity voxels from the determining step are clustered or distributed in the regions of interest;

evaluating whether there is a likelihood the heart has a global injury using the measures of intensity and spatial coordinate data of the high intensity voxels based on data from the identifying step to determine whether the patient has a change in tissue composition and/or function that is in a substantially randomly distributed pattern and/or in a pattern that is not detectable at a resolution of image slices associated with the medical image; and

providing output of the determined likelihood to a display associated with a clinician workstation.

2. A method according to claim 1, wherein the regions of interest include regions of interest derived from each of an apical, middle and basal portion of the heart.

3. A method according to claim 1, wherein the electronically identifying step comprises generating mean intensity voxel histograms, and wherein the evaluating step considers a characterizing predictive portion of the histogram that is associated with voxels having a mean intensity that is greater than 1 sigma standard deviation and less than about 3 sigma standard deviation and the three-dimensional location of those voxels.

4. A method according to claim 1, wherein the evaluating step comprises electronically assessing whether there is a clustering of voxels having an intensity that is in a range of between about 2-3 sigma standard deviation in histograms of voxel intensity whereby the 2-3 sigma standard deviation range defines a characterizing portion of the histogram.

5. A method according to claim 4, further comprising electronically identifying injury to cells in the heart prior to cell death, based on a characterizing portion of at least one of the histograms.

6. A method according to claim 1, further comprising automatically determining whether the voxels in the regions of interest are associated with an increased risk of cardiac injury associated with a decrease in heart function based on the evaluating step.

7. A method according to claim 1, wherein the evaluating step evaluates cardiac injury associated with drug-induced cardiotoxicity.

8. A method according to claim 1, further comprising generating data for a clinician to evaluate a cancer treatment therapy based on the evaluating step.

9. A method according to claim 8, wherein the generating data comprises generating data to alter, stop or decrease dosage of a chemotherapeutic agent for the patient to reduce cardiotoxicity to the patient when there is a likelihood of cardiac injury based on the evaluating step.

10. A method according to claim 1, wherein the evaluating step is carried out before a clinically relevant decrease in LVEF to thereby provide an early marker for a patient at-risk of developing irreversible cardiac injury.

11. A method according to claim 1, wherein the patient is an oncology patient, the method further comprising electronically predicting a future LVEF deterioration in the patient in response to chemotherapy.

12. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

electronically determining, for a plurality of regions of interest in a medical image of a heart, a location of voxels in three-dimensional space and an associated respective measure of intensity for each of the voxels;

electronically generating a plurality of histograms of voxel intensity, one for each region of interest corresponding to a CMR image slice of cardiovascular tissue of the patient, the histogram having a characterizing predictive portion; and

electronically evaluating whether there is a likelihood of present or future cardiac injury based on data from the generating step,

wherein the evaluating step comprises evaluating the characterizing predictive portion of the histogram.

13. A method according to claim 12,

wherein the evaluating step comprises evaluating a line shape of the characterizing portion of the histogram, the line shape of the characterizing portion being correlated to an increased risk of cardiac injury associated with chemotherapy-induced toxicity.

14. A method according to claim 12, wherein the histogram is a histogram of a distribution of a mean intensity of voxels/pixels.

15. A method according to claim 14, wherein the regions of interest include a plurality associated with a left ventricle myocardium.

16. A method according to claim 12, wherein the characterizing portion is a tail portion of the histogram.

17. A method according to claim 16, wherein the generating data comprises generating data allowing a clinician to increase dosage of a chemotherapeutic agent when the evaluating step indicates an absence of myocellular injury.

18. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

electronically determining, for a plurality of regions of interest in a medical image of a heart, a location of voxels in three-dimensional space and an associated respective measure of intensity for each of the voxels;

electronically evaluating whether there is a likelihood of present or future cardiac injury based on data from the determining step;

obtaining at least one cardiovascular magnetic resonance 3-D (CMR) image to generate voxel data used in the determining and evaluating steps; and

administering a chemotherapeutic agent before obtaining the at least one CMR image that is used to generate the voxel data, wherein the evaluating step is carried out to predict the likelihood of an undesired decrease in left ventricular ejection fraction (LVEF) based on a patients cardiotoxicity reaction to the administered chemotherapeutic agent.

19. A method according to claim 18, further comprising:

administering a contrast agent before obtaining the at least one CMR image; and

visually illustrating, in a compartmental model of a heart, randomly distributed regions of increased contrast uptake throughout at least a myocardium of a left ventricle.

20. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

electronically determining, for a plurality of regions of interest in a medical image of a heart, a location of voxels in three-dimensional space and an associated respective measure of intensity for each of the voxels;

electronically evaluating whether there is a likelihood of present or future cardiac injury based on data from the determining step;

obtaining at least one cardiovascular magnetic resonance (CMR) image for the determining step medical image;

electronically generating a plurality of histograms of voxel mean intensity using voxel data from the determining step, one for each region of interest, the histograms having a tail characterizing portion; and

wherein the evaluating step comprises automatically monitoring the patient'"'"'s cardiotoxicity response to chemotherapy using the characterizing portion of the histogram.

21. A method according to claim 20, wherein the characterizing portion of the histogram is associated with pixels/voxels in a region of interest that is based on substantially all of the pixels/voxels in the region of interest.

22. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

obtaining at least one CMR image that includes a left ventricle myocardium;

electronically determining, for a plurality of regions of interest in the CMR image, a location of voxels in three-dimensional space and an associated respective measure of intensity for each of the voxels, wherein the determining step includes electronically interrogating at least substantially all voxels within a contiguous series of short axis slice positions spanning an apex to a base in the at least one CMR image and identifying the intensity and x, y and z coordinate of each voxel in three-dimensional space;

electronically generating a non-invasive biopsy image of the patient that can identify locations of voxels of similar intensities within the left ventricle using data from the determining step; and

evaluating whether there is a likelihood of present or future cardiac injury based on data from the determining step.

23. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

obtaining at least one CMR image that includes a right ventricle myocardium;

electronically determining, for a plurality of regions of interest in the CMR image, a location of voxels in three-dimensional space and an associated respective measure of intensity for each of the voxels, wherein the determining step includes electronically interrogating at least substantially all voxels within a contiguous series of short axis slice positions spanning an apex to a base in the at least one CMR image and identifying the intensity and x, y and z coordinate of each voxel in three-dimensional space;

electronically generating a non-invasive biopsy image of the patient that can identify locations of voxels of similar intensities within the right ventricle using data from the determining step; and

evaluating whether there is a likelihood of present or future cardiac injury based on data from the determining step.

24. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

obtaining at least one CMR image;

electronically interrogating at least substantially all voxels within a contiguous series of short axis slice positions spanning an apex to a base in the at least one CMR image and identifying an intensity and x, y and z coordinate of each voxel in three-dimensional space; and

electronically generating a non-invasive biopsy image of the patient'"'"'s heart that can identify locations of voxels of similar intensities within the heart for use in evaluating whether there is a likelihood of present or future cardiac injury based on data from the interrogating step.

25. A method according to claim 24, further comprising electronically adjusting the intensity of the voxels to account for background noise.

26. A non-invasive method of evaluating actual and/or potential cardiac injury in a patient, comprising:

obtaining at least one Magnetic Resonance Imaging (MRI) image and/or X-ray Computed Tomography image of a heart of a patient;

electronically determining, for a plurality of regions of interest in the obtained image of the heart, a location of voxels in three-dimensional space and an associated respective measure of intensity for each of the voxels;

electronically extracting and storing x, y, and z coordinates of voxel intensity data of each voxel in a myocardium region of the at least one image;

electronically adjusting voxel intensity data for background noise in the myocardium region;

generating histograms of the regions of interest based on the determining step using data from the extracting and storing step; and

electronically evaluating whether there is a likelihood of present or future cardiac injury based on data from the generating histogram step.

27. A method of predicting cardiac injury prior to an irreversible dysfunctional state, comprising:

electronically extracting and storing x, y, and z coordinates of voxel intensity data of voxels in multiple slices of an MRI image of a left ventricle myocardium;

electronically generating at least one histogram of mean intensities of voxels/pixels using the extracted and stored voxel data;

electronically determining a likelihood of cardiac injury due to cardiotoxicity based on data from the at least one histogram; and

providing output of the determined likelihood to a display associated with a clinician workstation.

28. A method according to claim 27, wherein the electronically determining step comprises distinguishing cardiac injury associated with myocardial infarct from cardiac injury associated with cardiotoxicity using data from a characterizing predictive portion of at least one histogram.

29. A method according to claim 27, wherein the electronically determining step comprises evaluating a likelihood of a future decrement in LVEF due to cardiotoxicity based on data from the histogram.

30. A method according to claim 27, wherein the determining step comprises electronically evaluating a tail portion of the histogram as a predictor of cardiotoxicity.

31. A method according to claim 27, wherein the at least one histogram is a plurality of histograms, and wherein the determining step comprises electronically evaluating the histograms to determine locations of clusters of voxels of similar intensity.

32. A signal processor circuit comprising:

a signal processor configured to determine a likelihood of cardiac injury of a patient'"'"'s heart due to cardiotoxicity using histograms of mean intensity of voxels from multiple different slices of an MRI or X-ray CT image of cardiac tissue, wherein the histograms represent percentage versus mean intensity of voxels within a region of interest, and wherein the signal processor is configured to evaluate whether the patient has a global cardiac injury using measures of mean intensity and associated spatial coordinate data of the voxels to thereby determine whether the patient has a change in tissue composition and/or function that is in a substantially randomly distributed pattern and/or in a pattern that is not detectable at a resolution of the image slices.

33. A signal processor circuit according to claim 32, wherein the signal processor is configured to evaluate a characterizing portion, distribution pattern or lineshape of the histograms to determine the likelihood of cardiac injury due to cardiotoxicity.

34. A signal processor circuit according to claim 32, wherein the signal processor is configured to distinguish cardiac injury associated with myocardial infarct from cardiac injury associated with cardiotoxicity based on a shape of a tail portion of the histograms.

35. A signal processor circuit according to claim 32, wherein the signal processor is configured to determine a likelihood of a decrement in LVEF due to cardiotoxicity based on data from at least one of the histograms to thereby allow alternative treatment protocols before irreversible cardiac injury occurs.

36. A signal processor circuit according to claim 32, wherein the signal processor is configured to evaluate a shape, lineshape or distribution pattern of a tail portion of the histogram as a predictor of cardiac injury due to cardiotoxicity.

37. A signal processor circuit according to claim 36, wherein the signal processor is configured to determine where clusters of voxels with similar intensities are located in the patients heart.

38. A non-invasive system for evaluating cardiotoxicity, comprising:

a signal processor in communication with a physician workstation configured to generate at least one histogram of intensity voxels of at least one image of cardiac tissue of a patient and determine the likelihood of cardiac injury due to cardiotoxicity using a characterizing portion, distribution pattern or lineshape of the at least one histogram,

wherein the intensity voxels are associated with an MRI or X-ray CT image of cardiac tissue, and wherein the histogram represents percentage versus a measure of intensity of voxels within a region of interest, and wherein the signal processor is configured to evaluate whether the cardiac tissue has a global injury using measures of intensity and spatial coordinate data of the voxels to thereby determine whether the patient has a change in tissue composition and/or function that is in a substantially randomly distributed pattern and/or in a pattern that is not detectable at a resolution of image slices associated with the image of cardiac tissue.

39. A system according to claim 38, wherein the at least one histogram comprises a mean intensity histogram.

40. A system according to claim 39, wherein a plurality of mean intensity histograms are generated, including histograms for a plurality of regions of interest for slice voxels associated with different axial positions of the left ventricle.

41. A signal processor circuit according to claim 38, wherein the signal processor is configured to distinguish cardiac injury associated with myocardial infarct from cardiac injury associated with cardiotoxicity.

42. A signal processor circuit according to claim 38, wherein the signal processor is configured to determine a likelihood of a decrement in LVEF due to cardiotoxicity based on data from the histogram to thereby allow alternative oncology treatment protocols before irreversible cardiac injury occurs.

43. A signal processor circuit according to claim 38, wherein the signal processor is configured to evaluate a shape, lineshape or distribution pattern of a tail portion of the histogram as a predictor of cardiac injury due to cardiotoxicity.

44. A computer program product for evaluating cardiac injury a patient'"'"'s heart, comprising:

a computer readable medium having computer readable program code embodied therein, the computer readable program code comprising;

computer readable program code configured to determine a likelihood of cardiac injury associated with cardiotoxicity using a characterizing portion of at least one histogram of a measure of intensity of voxels in at least one cardiac image obtained after administration of a contrast agent to the patient, wherein the computer readable program code is configured to evaluate whether the cardiac injury is a global injury using the voxel measures of intensity and associated spatial coordinate data to thereby determine whether the patient has a change in tissue composition and/or function that is in a substantially randomly distributed pattern and/or in a pattern that is not detectable at a resolution of image slices associated with the at least one cardiac image.

45. A computer program product according to claim 44, further comprising computer readable program code configured to correlate voxel intensity values with x, y and z coordinates, and wherein the voxel intensity values are taken from least one CMR (Cardio Magnetic Resonance) image.

46. A computer program product according to claim 44, further comprising:

computer readable program code configured to obtain a first image of a region of interest outside the heart corresponding to the first cardiac image;

computer readable program code configured to correct for variations in intensity of voxels in the at least one cardiac image using the first image of a region of interest outside the heart.

47. A computer program product according to claim 44, wherein the computer readable program code configured to determine a likelihood of cardiac injury associated with cardiotoxicity is configured to distinguish cardiac injury associated with myocardial infarct from cardiac injury associated with cardiotoxicity.

48. A computer program product according to claim 44, wherein the computer readable program code configured to determine a likelihood of cardiac injury associated with cardiotoxicity comprises computer readable program code configured to determine a likelihood of a decrement in LVEF due to cardiotoxicity based on data from the histogram to thereby allow alternative oncology treatment protocols before irreversible cardiac injury occurs.

49. A computer program product according to claim 44, wherein the computer readable program code configured to determine a likelihood of cardiac injury associated with cardiotoxicity is configured to evaluate a shape, lineshape or distribution pattern of a tail portion of the histogram as a characterizing portion that is predictive of cardiac injury due to cardiotoxicity.

50. A system for non-invasively predicting cardiac injury due to cardiotoxicity prior to an irreversible state of cardiac injury associated with clinical dysfunction:

a signal processor circuit in communication with a display at a physician workstation, the signal processor configured to electronically generate at least one histogram of mean intensity of voxels/pixels in an MRI or CT image and electronically determine a likelihood of cardiac global injury of a patient due to cardiotoxicity based on data from the at least one histogram using measures of intensity and spatial coordinate data to thereby determine whether the patient has a change in tissue composition and/or function that is in a substantially randomly distributed pattern and/or in a pattern that is not detectable at a resolution of image slices associated with the MRI or CT image.

51. A system according to claim 50, wherein the signal processor circuit comprises a statistical model that is configured to distinguish cardiac injury associated with myocardial infarct from cardiac injury associated with cardiotoxicity.

52. A system according to claim 50, wherein the signal processor circuit that is configured to electronically determine a likelihood of cardiac injury is configured to evaluate a likelihood of a future decrement in LVEF due to cardiotoxicity based on data from the histogram.

53. A system according to claim 50, wherein the signal processor circuit for electronically determining cardiac injury is configured to evaluate a tail portion of the histogram to determine the likelihood of cardiac injury due to cardiotoxicity.

54. A system according to claim 50, wherein the at least one histogram is a plurality of histograms, including a plurality of histograms for regions of interest in a left ventricle.

55. A system according to claim 50, wherein the at least one histogram is a plurality of histograms, including a plurality of histograms for regions of interest in a right ventricle.

56. A system according to claim 50, wherein the at least one histogram is a plurality of histograms, one for each of 17 different defined compartments of the heart as defined in a standardized heart model.

57. A system for non-invasively predicting injury to an organ due to toxicity:

a clinician workstation with a display; and

a signal processor circuit in communication with the display, the signal processor configured to;

(a) identify measures of intensity of voxels of a target organ and correlate associated x, y and z coordinates of the voxels in three-dimensional space using multiple image slices of at least one imaging modality; and

(b) evaluate whether the target organ has a global injury using the measures of intensity and spatial coordinate data to thereby determine whether the patient has a change in tissue composition and/or function that is in a substantially randomly distributed pattern and/or in a pattern that is not detectable at a resolution of the image slices.

58. A system according to claim 57, wherein the signal processor is configured to render a 3-D image of the organ using the identified measures of voxel intensity and correlated spatial coordinates, wherein the 3-D image visually illustrates different values of quantified voxel intensities differently so that voxels having similar measures of intensities have similar greyscale or color to visually emphasize a distribution of injured or dead cells over the organ for ease of viewability and reference for a clinician.

59. A system according to claim 57, wherein the image visually illustrates spatial locations of clusters of quantified measures of intensity of voxels having similar intensity measures, wherein clusters of similar high or low intensity pixels/voxels scattered about substantially an entire organ indicates a likelihood of present global injury or a risk of future global injury.

60. A system according to claim 57, wherein the target organ is the heart, and wherein the signal processor is configured to generate an image that illustrates randomly distributed clusters of mid-level intensity pixels/voxels in defined compartments of the heart.