Method and device for retinal image analysis

1. A method for diagnosing and/or predicting the presence, progression and/or treatment effect of a disease characterized by retinal pathological changes in a subject, comprising the steps of:

(a) obtaining a retinal fundus image from the subject;

(b) converting the image to a gray level image and/or extracting vessels from the image to obtain the gray level image and/or the vessel image;

(c) performing statistical texture analysis on the gray level image and/or the vessel image, and generating one or more factors, wherein the statistical texture analysis includes using gray level co-occurrence matrix (GLCM) and/or run length matrix (RLM) to obtain texture features, including normalized Homogeneity, normalized Entropy, normalized Contrast, normalized 4 th moment and normalized Run percentage;

(d) comparing the one or more factors obtained from step (c) with those of a control, a change of the factors is an indication of the presence, progression and/or treatment effect of the disease in the subject.

2. The method of claim 1 , wherein the disease is selected from the group consisting of stroke, hypertension, diabetes, cardiovascular diseases including coronary heart disease and cerebral vascular disease, glaucoma, prematurity, papilloedema, and common retina disease.

3. The method of claim 1 , wherein the disease is stroke, and step (c) includes performing the statistical texture analysis and at least one of fractal analysis, high order spectra analysis, and abnormal pattern analysis on the gray level image and/or the vessel image, and generating one or more factors.

4. The method of claim 1 , wherein the disease is diabetes, and step (c) includes performing the statistical texture analysis and at least one of fractal analysis, high order spectra analysis, and abnormal pattern analysis on the gray level image and/or the vessel image, and generating one or more factors.

5. The method of claim 1 , wherein step (b) includes using mathematical morphological operations and/or dual tree complex wavelet transform technique.

6. The method of claim 1 , wherein step (c) further comprises performing one or two or three analyses selected from the group consisting of fractal analysis, high order spectra analysis, and abnormal pattern analysis on the gray level image and/or the vessel image, and generating one or more factors.

7. The method of claim 6 , wherein the fractal analysis in step (c) includes using complex wavelet methods.

8. The method of claim 6 , wherein the fractal analysis in step (c) includes using multifractal spectrum analysis.

9. The method of claim 6 , wherein the factor generated by the fractal analysis is selected from the group consisting of Sum Average, Cluster Shade, Hausdorff dimension, and any combination thereof.

10. The method of claim 6 , wherein the higher order spectral (HOS) features obtained from the high order spectra analysis in step (c) include Entropy1 HOS features at degree of 100, Entropy2 HOS features at degree of 160, Entropy3 HOS features at degree of 40, 140 and 160; Entropy Phase HOS features at degree of 40, 140, 160 and 180; Entropy Magnitude HOS features at degree of 0, 20, 40, 60, 100, 120, 140, 160 and 180.

11. The method of claim 6 , wherein the abnormal pattern analysis in step (c) includes using a wavelet algorithm based on wavelet transform.

12. The method of claim 6 , wherein the abnormal pattern analysis in step (c) includes using higher order spectral analysis in combination with a wavelet algorithm based on wavelet transform.

13. The method of claim 6 , wherein the abnormal patterns include hemorrhages, exudates, new vessels, microaneurysm, proliferative vitreoretinopathy or any combination thereof.

14. The method of claim 13 , wherein the abnormal patterns are hemorrhages.

15. A method for generating one or more factors associated with retinal pathological changes in a disease of a subject, comprising the steps of:

(a) obtaining a retinal fundus image from the subject;

(b) converting the image to a gray level image and/or extracting vessels from the image to obtain the gray level image and/or the vessel image; and

(c) performing statistical texture analysis on the gray level image and/or the vessel image, and generating one or more factors, wherein the statistical texture analysis includes using gray level co-occurrence matrix (GLCM) and/or run length matrix (RLM) to obtain texture features, including normalized Homogeneity, normalized Entropy, normalized Contrast, normalized 4 th moment and normalized Run percentage.

16. The method of claim 15 , wherein the disease is selected from the group consisting of stroke, hypertension, diabetes, cardiovascular diseases including coronary heart disease and cerebral vascular disease, glaucoma, prematurity, papilloedema, and common retina disease.

17. The method of claim 15 , wherein the disease is stroke, and step (c) includes performing the statistical texture analysis and at least one of fractal analysis, high order spectra analysis, and abnormal pattern analysis on the gray level image and/or the vessel image, and generating one or more factors.

18. The method of claim 15 , wherein the disease is diabetes, and step (c) includes performing the statistical texture analysis and at least one of fractal analysis, high order spectra analysis, and abnormal pattern analysis on the gray level image and/or the vessel image, and generating one or more factors.

19. The method of claim 15 , wherein step (b) includes using mathematical morphological operations and/or dual tree complex wavelet transform technique.

20. The method of claim 15 , wherein step (c) further comprises performing one or two or three analyses selected from the group consisting of fractal analysis, high order spectra analysis, and abnormal pattern analysis on the gray level image and/or the vessel image, and generating one or more factors.

21. The method of claim 20 , wherein the fractal analysis in step (c) includes using complex wavelet methods.

22. The method of claim 20 , wherein the fractal analysis in step (c) includes using multifractal spectrum analysis.

23. The method of claim 20 , wherein the factor generated by the fractal analysis is selected from the group consisting of Sum Average, Cluster Shade, Hausdorff dimension, and any combination thereof.

24. The method of claim 20 , wherein the higher order spectral (HOS) features obtained from the high order spectra analysis in step (c) include Entropy1 HOS features at degree of 100, Entropy2 HOS features at degree of 160, Entropy3 HOS features at degree of 40, 140 and 160; Entropy Phase HOS features at degree of 40, 140, 160 and 180; Entropy Magnitude HOS features at degree of 0, 20, 40, 60, 100, 120, 140, 160 and 180.

25. The method of claim 20 , wherein the abnormal pattern analysis in step (c) includes using a wavelet algorithm based on wavelet transform.

26. The method of claim 20 , wherein the abnormal pattern analysis in step (c) includes using higher order spectral analysis in combination with a wavelet algorithm based on wavelet transform.

27. The method of claim 20 , wherein the abnormal patterns include hemorrhages, exudates, new vessels, microaneurysm, proliferative vitreoretinopathy or any combination thereof.

28. The method of claim 27 , wherein the abnormal patterns are hemorrhages.

29. A device for diagnosing and/or predicting the presence, progression and/or treatment effect of a disease characterized by retinal pathological changes in a subject, comprising:

an image-capturing module for obtaining a retinal fundus image from the subject;

a conversion module for converting the image to a gray level image and/or extracting vessels from the image to obtain said gray level image and/or vessel image; and

an analysis module comprising a statistical texture analysis submodule for performing analysis on the gray level image and/or vessel image and generating one or more factors, wherein the statistical texture analysis submodule is configured to carry out gray level co-occurrence matrix (GLCM) and/or run length matrix (RLM) to obtain texture features, including normalized Homogeneity, normalized Entropy, normalized Contrast, normalized 4 th moment and normalized Run percentage.

30. The device of claim 29 , wherein the disease is selected from the group consisting of stroke, hypertension, diabetes, cardiovascular diseases including coronary heart disease and cerebral vascular disease, glaucoma, prematurity, papilloedema, and common retina disease.

31. The device of claim 29 , wherein the disease is stroke, and the analysis module comprises the statistical texture analysis submodule and at least one of a fractal analysis submodule, a high order spectra analysis submodule, and an abnormal pattern analysis submodule for performing analysis on the gray level image and/or vessel image and generating one or more factors.

32. The device of claim 29 , wherein the disease is diabetes, and the analysis module comprises the statistical texture analysis submodule and at least one of a fractal analysis submodule, a high order spectra analysis submodule, and an abnormal pattern analysis submodule for performing analysis on the gray level image and/or vessel image and generating one or more factors.

33. The device of claim 29 , wherein the conversion module is configured to carry out mathematical morphological operations and/or dual tree complex wavelet transform technique.

34. The device of claim 29 , wherein the analysis module further comprises one or two or three submodules selected from the group consisting of a fractal analysis submodule, a high order spectra analysis submodule, and an abnormal pattern analysis submodule for performing analysis on the gray level image and/or vessel image and generating one or more factors.

35. The device of claim 34 , wherein the fractal analysis submodule is configured to carry out complex wavelet methods.

36. The device of claim 34 , wherein the fractal analysis submodule is configured to carry out multifractal spectrum analysis.

37. The device of claim 34 , wherein the factor generated by the fractal analysis submodule is selected from the group consisting of Sum Average, Cluster Shade, Hausdorff dimension, and any combination thereof.

38. The device of claim 34 , wherein the higher order spectral (HOS) features generated by the high order spectra analysis submodule include Entropy1 HOS features at degree of 100, Entropy2 HOS features at degree of 160, Entropy3 HOS features at degree of 40, 140 and 160; Entropy Phase HOS features at degree of 40, 140, 160 and 180; Entropy Magnitude HOS features at degree of 0, 20, 40, 60, 100, 120, 140, 160 and 180.

39. The device of claim 34 , wherein the abnormal pattern analysis submodule is configured to carry out a wavelet algorithm based on wavelet transform.

40. The device of claim 34 , wherein the abnormal pattern analysis submodule is configured to carry out higher order spectral analysis in combination with a wavelet algorithm based on wavelet transform.

41. The device of claim 34 , wherein the abnormal patterns analyzed by the abnormal pattern analysis submodule include hemorrhages, exudates, new vessels, microaneurysm, proliferative vitreoretinopathy or any combination thereof.

42. The device of any of claim 41 , wherein the abnormal patterns are hemorrhages.

43. The method of claim 2 , wherein the common retina disease is macular hole or age-related macular degeneration.

44. The method of claim 7 , wherein the fractal analysis in step (c) includes using complex wavelet Leader multifractal analysis.

45. The method of claim 11 , wherein the abnormal pattern analysis in step (c) includes using a dual tree complex wavelet transform and a wavelet-based Radon transform.

46. The method of claim 12 , wherein the wavelet algorithm based on wavelet transform is a dual tree complex wavelet transform and a wavelet-based Radon transform.

47. The method of claim 16 , wherein the common retina disease is macular hole or age-related macular degeneration.

48. The method of claim 21 , wherein the fractal analysis in step (c) includes using complex wavelet Leader multifractal analysis.

49. The method of claim 25 , wherein the abnormal pattern analysis in step (c) includes using a dual tree complex wavelet transform and a wavelet-based Radon transform.

50. The method of claim 26 , wherein the wavelet algorithm based on wavelet transform is a dual tree complex wavelet transform and a wavelet-based Radon transform.

51. The device of claim 29 , further comprising a comparison module for comparing the factors obtained from the analysis module with those of a control.

52. The device of claim 30 , wherein the common retina disease is macular hole or age-related macular degeneration.

53. The device of claim 35 , wherein the fractal analysis submodule is configured to carry out complex wavelet Leader multifractal analysis.

54. The device of claim 39 , wherein the abnormal pattern analysis submodule is configured to carry out a dual tree complex wavelet transform and a wavelet-based Radon transform.

55. The device of claim 40 , wherein the wavelet algorithm based on wavelet transform is a dual tree complex wavelet transform and a wavelet-based Radon transform.