Method and apparatus for automated detection of target structures from medical images using a 3D morphological matching algorithm

• US 7,949,169 B2
• Filed: 12/03/2003
• Issued: 05/24/2011
• Est. Priority Date: 12/04/2002
• Status: Expired due to Fees

• Current Assignee: Jinsung Kim, Kyongtae T. Bae
• Original Assignee: Jinsung Kim, Kyongtae T. Bae
• Inventors: Kim, Jinsung, Bae, Kyongtae T.
• Primary Examiner(s): TUCKER, WESLEY J
• Application Number:
  US10/537,514
• Publication Number:
  US 20070140541A1
• Time in Patent Office: 2,729 Days
• Field of Search: 382128-134, 382/154
• US Class Current: 382/131
• CPC Class Codes: G06T 2207/30061 Lung
• CPC Class Codes: G06T 2207/30101 Blood vessel; Artery; Vein; Vascular
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

A method for the automated detection of target structures shown in digital medical images, the method of comprising: (1) generating a three dimensional (3D) volumetric data set of a patient region within which the target structure resides from a plurality of segmented medical image slices; (2) grouping contiguous structures that are depicted in the 3D volumetric data set to create corresponding grouped structure data sets; (3) assigning each grouped structure data set to one of a plurality of detection algorithms, each detection algorithm being configured to detect a different type of target structure; and (4) processing each grouped structure data set according to its assigned detection algorithm to thereby detect whether any target structures are present in the medical images. Preferably, the target structures are pulmonary nodules, and a specialized detection algorithm is applied to image data classified as a candidate for depicting perivascular nodules. To segment perivascular nodule candidates from surrounding vessels, the image data is preferably correlated with a plurality of 3D morphological filters.

Claims

1. A device for detecting whether pulmonary nodules are present in a patient'"'"'s lung region from a three-dimensional (3D) data set representative of a volumetric image of the patient'"'"'s lung region, the device comprising:

a processor configured to (1) identify contiguous structures in the 3D data set, (2) classify the identified contiguous structures according to a plurality of classifications, the classifications comprising a vessel contiguous structure classification and a non-vessel contiguous structure classification, wherein each non-vessel contiguous structure comprises a nodule candidate, (3) apply a first nodule detection operation to each vessel contiguous structure to determine a nodule status therefor, and (4) apply a second nodule detection operation to each non-vessel contiguous structure to determine a nodule status therefor, wherein the first nodule detection operation is different than the second nodule detection algorithm;

wherein the processor is further configured to apply the first nodule detection operation by (1) segmenting nodule candidate structures from surrounding vessel structures through a correlation of each vessel contiguous structure with a 3D morphological filter, and (2) determining a nodule status for each segmented nodule candidate;

wherein the processor is further configured to apply the second nodule detection operation by determining a nodule status for each non-vessel nodule candidate at least partially on the basis of geometric criteria;

wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with a compactness criteria; and

wherein the processor is further configured to compare each non-vessel nodule candidate with the compactness criteria by;

for each non-vessel nodule candidate, (1) determining a volume of that non-vessel nodule candidate, (2) determining a volume of the smallest 3D box that encloses that non-vessel nodule candidate, and (3) computing a ratio of the determined non-vessel nodule candidate volume to the determined box volume; and

for each non-vessel nodule candidate having a determined volume ratio less than approximately 0.5 or greater than approximately 1.5, determining that that non-vessel nodule candidate is not a pulmonary nodule.

2. The device of claim 1 wherein the processor is further configured to perform the correlation by convolving each vessel contiguous structure with a 3D morphological filter.

3. The device of claim 2 wherein the processor is further configured to perform the correlation by convolving each vessel contiguous structure with a plurality of 3D morphological filters.

4. The device of claim 3 wherein the filters comprise a plurality of spherical filters, each filter being tuned with a different diameter.

5. The device of claim 3 wherein the processor is further configured to determine nodule status for the segmented nodule candidates by determining a nodule status for each nodule candidate at least partially on the basis of geometric criteria.

6. The device of claim 5 wherein the geometric criteria includes size.

7. The device of claim 5 wherein the geometric criteria includes compactness.

8. The device of claim 5 wherein the geometric criteria includes elongation.

9. The device of claim 5 wherein the geometric criteria includes size, compactness, and elongation.

10. The device of claim 3 wherein the processor is further configured to perform the segmentation by determining that a nodule candidate exists if the correlation results in a correlation within a predetermined range of correlation values.

11. The device of claim 1 wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with a size criteria.

12. The device of claim 11 wherein the size criteria is diameter.

13. The device of claim 1 wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with an elongation criteria.

14. The device of claim 13 wherein the processor is further configured to compare each non-vessel nodule candidate with an elongation criteria by comparing each non-vessel nodule candidate with a two-dimensional (2D) elongation criteria and a 3D elongation criteria.

15. The device of claim 14 wherein the processor is further configured to compare each non-vessel nodule candidate with an elongation criteria by comparing a non-vessel nodule candidate with the 3D elongation criteria only if that non-vessel nodule candidate satisfies the 2D elongation criteria.

16. The device of claim 1 wherein the processor is further configured to determine a nodule status for each non-vessel nodule candidate by comparing each nodule candidate with a size criteria, a compactness criteria, and an elongation criteria.

17. The device of claim 1 wherein the processor is further configured to generate the 3D data set from a plurality of 2D image slices of the patient'"'"'s lung region.

18. The device of claim 17 wherein the 2D image slices comprise a plurality of computed tomography slices.

19. The device of claim 17 wherein the 2D image slices comprise a plurality of magnetic resonance slices.

20. The device of claim 17 wherein the 2D image slices comprise a plurality of ultrasound slices.

21. A device for detecting whether pulmonary nodules are present in a patient'"'"'s lung region from a three-dimensional (3D) data set representative of a volumetric image of the patient'"'"'s lung region, the device comprising:

a processor configured to (1) identify contiguous structures in the 3D data set, (2) classify the identified contiguous structures according to a plurality of classifications, the classifications comprising a vessel contiguous structure classification and a non-vessel contiguous structure classification, wherein each non-vessel contiguous structure comprises a nodule candidate, (3) apply a first nodule detection operation to each vessel contiguous structure to determine a nodule status therefor, and (4) apply a second nodule detection operation to each non-vessel contiguous structure to determine a nodule status therefor, wherein the first nodule detection operation is different than the second nodule detection algorithm;

wherein the processor is further configured to apply the first nodule detection operation by (1) segmenting nodule candidate structures from surrounding vessel structures through a correlation of each vessel contiguous structure with a 3D morphological filter, and (2) determining a nodule status for each segmented nodule candidate;

wherein the processor is further configured to apply the second nodule detection operation by determining a nodule status for each non-vessel nodule candidate at least partially on the basis of geometric criteria;

wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with an elongation criteria; and

wherein the processor is further configured to compare each non-vessel nodule candidate with the elongation criteria by;

for each non-vessel nodule candidate, (1) determining a length of a major axis of the smallest rectangle or ellipse that encloses that non-vessel nodule candidate, (2) determining a length of a minor axis of the smallest rectangle or ellipse that encloses that non-vessel nodule candidate, (3) computing a ratio of the major axis length to the minor axis length; and

for each non-vessel nodule candidate having a determined elongation axis ratio greater than approximately 3.0, determining that that non-vessel nodule candidate is not a pulmonary nodule.

22. The device of claim 21 wherein the processor is further configured to perform the correlation by convolving each vessel contiguous structure with a 3D morphological filter.

23. The device of claim 22 wherein the processor is further configured to perform the correlation by convolving each vessel contiguous structure with a plurality of 3D morphological filters.

24. The device of claim 23 wherein the filters comprise a plurality of spherical filters, each filter being tuned with a different diameter.

25. The device of claim 23 wherein the processor is further configured to determine nodule status for the segmented nodule candidates by determining a nodule status for each nodule candidate at least partially on the basis of geometric criteria.

26. The device of claim 25 wherein the geometric criteria includes size.

27. The device of claim 25 wherein the geometric criteria includes compactness.

28. The device of claim 25 wherein the geometric criteria includes size, compactness, and elongation.

29. The device of claim 23 wherein the processor is further configured to perform the segmentation by determining that a nodule candidate exists if the correlation results in a correlation within a predetermined range of correlation values.

30. The device of claim 21 wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with a size criteria.

31. The device of claim 30 wherein the size criteria is diameter.

32. The device of claim 21 wherein the processor is further. configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with the elongation criteria.

33. The device of claim 32 wherein the processor is further configured to compare each non-vessel nodule candidate with an elongation criteria by comparing each non-vessel nodule candidate with a two-dimensional (2D) elongation criteria and a 3D elongation criteria.

34. The device of claim 33 wherein the processor is further configured to compare each non-vessel nodule candidate with an elongation criteria by comparing a non-vessel nodule candidate with the 3D elongation criteria only if that non-vessel nodule candidate satisfies the 2D elongation criteria.

35. The device of claim 21 wherein the processor is further configured to generate the 3D data set from a plurality of 2D image slices of the patient'"'"'s lung region.

36. The device of claim 35 wherein the 2D image slices comprise a plurality of computed tomography slices.

37. The device of claim 35 wherein the 2D image slices comprise a plurality of magnetic resonance slices.

38. The device of claim 35 wherein the 2D image slices comprise a plurality of ultrasound slices.

39. A device for detecting whether pulmonary nodules are present in a patient'"'"'s lung region from a three-dimensional (3D) data set representative of a volumetric image of the patient'"'"'s lung region, the device comprising:

a processor configured to (1) identify contiguous structures in the 3D data set, (2) classify the identified contiguous structures according to a plurality of classifications, the classifications comprising a vessel contiguous structure classification and a non-vessel contiguous structure classification, wherein each non-vessel contiguous structure comprises a nodule candidate, (3) apply a first nodule detection operation to each vessel contiguous structure to determine a nodule status therefor, and (4) apply a second nodule detection operation to each non-vessel contiguous structure to determine a nodule status therefor, wherein the first nodule detection operation is different than the second nodule detection algorithm;

wherein the processor is further configured to apply the first nodule detection operation by (1) segmenting nodule candidate structures from surrounding vessel structures through a correlation of each vessel contiguous structure with a 3D morphological filter, and (2) determining a nodule status for each segmented nodule candidate

wherein the processor is further configured to apply the second nodule detection operation by determining a nodule status for each non-vessel nodule candidate at least partially on the basis of geometric criteria;

wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with an elongation criteria; and

wherein the processor is further configured to compare each non-vessel nodule candidate with the elongation criteria by;

for each non-vessel nodule candidate, (1) determining a maximum eigenvalue from coordinates of the voxels of that non-vessel nodule candidate, (2) determining a minimum eigenvalue from coordinates of the voxels of that non-vessel nodule candidate, (3) computing a ratio of the maximum eigenvalue to the minimum eigenvalue; and

for each non-vessel nodule candidate having a determined elongation eigenvalue ratio greater than approximately 3.0, determining that that non-vessel nodule candidate is not a pulmonary nodule.

40. The device of claim 39 wherein the processor is further configured to perform the correlation by convolving each vessel contiguous structure with a 3D morphological filter.

41. The device of claim 40 wherein the processor is further configured to perform the correlation by convolving each vessel contiguous structure with a plurality of 3D morphological filters.

42. The device of claim 41 wherein the filters comprise a plurality of spherical filters, each filter being tuned with a different diameter.

43. The device of claim 41 wherein the processor is further configured to determine nodule status for the segmented nodule candidates by determining a nodule status for each nodule candidate at least partially on the basis of geometric criteria.

44. The device of claim 43 wherein the geometric criteria includes size.

45. The device of claim 43 wherein the geometric criteria includes compactness.

46. The device of claim 43 wherein the geometric criteria includes size, compactness, and elongation.

47. The device of claim 41 wherein the processor is further configured to perform the segmentation by determining that a nodule candidate exists if the correlation results in a correlation within a predetermined range of correlation values.

48. The device of claim 39 wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with a size criteria.

49. The device of claim 48 wherein the size criteria is diameter.

50. The device of claim 39 wherein the processor is further configured to determine the nodule status for each non-vessel nodule candidate by comparing each nodule candidate with the elongation criteria.

51. The device of claim 50 wherein the processor is further configured to compare each non-vessel nodule candidate with an elongation criteria by comparing each non-vessel nodule candidate with a two-dimensional (2D) elongation criteria and a 3D elongation criteria.

52. The device of claim 51 wherein the processor is further configured to compare each non-vessel nodule candidate with an elongation criteria by comparing a non-vessel nodule candidate with the 3D elongation criteria only if that non-vessel nodule candidate satisfies the 2D elongation criteria.

53. The device of claim 39 wherein the processor is further configured to generate the 3D data set from a plurality of 2D image slices of the patient'"'"'s lung region.

54. The device of claim 53 wherein the 2D image slices comprise a plurality of computed tomography slices.

55. The device of claim 53 wherein the 2D image slices comprise a plurality of magnetic resonance slices.

56. The device of claim 53 wherein the 2D image slices comprise a plurality of ultrasound slices.

57. A device for analyzing a 3D data set representative of a patient'"'"'s lung region, the device comprising:

a processor configured to (1) group the data set into data subsets, each subset being representative of a contiguous structure, (2) identify each data subset that corresponds to a vessel, and (3) segment any perivascular nodule candidates from each identified subset by correlating that identified subset with at least one 3D morphological filter that is tuned to an expected shape of a perivascular nodule;

wherein the processor is further configured to perform the correlation by convolving each identified subset with the at least one filter to thereby compute a correlation value; and

wherein the processor is further configured to perform the convolution by computing the correlation value between the identified subset (I) and the filter (F) by a Fast Fourier Transform (FFT) according to the formula;

58. The device of claim 57 wherein the processor is configured to perform the correlation by correlating each identified subset with a plurality of 3D morphological filters, each filter being tuned to a different expected shape of a perivascular nodule.

59. The device of claim 58 wherein at least one filter is a spherical filter tuned with a predetermined diameter.

60. The device of claim 59 wherein the predetermined diameter is approximately 3 mm.

61. The device of claim 58 wherein a plurality of the filters are spherical filters, each spherical filter being tuned with a different predetermined diameter.

62. The device of claim 58 wherein the processor is further configured to perform the correlation by performing the correlations in parallel, each parallel correlation being configured to correlate an identified subset with one of the filters.

63. The device of claim 57 wherein the processor is further configured to perform the segmentation by determining whether a perivascular nodule candidate exists at least partially according to the computed correlation value.

64. The device of claim 63 wherein the processor is further configured to determine whether a perivascular nodule candidate exists by determining that a perivascular nodule candidate does exist if the computed correlation value lies within a predetermined range of correlation values.

65. The device of claim 57 wherein the processor is further configured to, for each segmented perivascular nodule candidate, determine a nodule status therefor at least partially on the basis of geometric criteria.

66. The device of claim 65 wherein the geometric criteria comprises at least one selected from the group consisting of candidate size, candidate compactness, and candidate elongation.

67. The device of claim 57 wherein the processor is further configured to generate the 3D data set from one selected from the group consisting of a plurality of computed tomography (CT) slices, a plurality of magnetic resonance (MR) slices, and a plurality of ultrasound slices.

68. The device of claim 57 wherein the processor is further configured to (1) identify each data subset that corresponds to a non-vessel, and (2) for each subset identified as corresponding to a non-vessel, determine a nodule status therefor at least partially on the basis of geometric criteria.