System and method for detecting solid components of ground glass nodules in pulmonary computed tomography images

1. A method for detecting solid components in ground glass nodules (GGNs) in medical images, comprising:

performing an intensity-based segmentation on a segmented GGN to identify a high intensity region; and

performing a shape analysis to determine whether the high intensity region is a solid component or a vessel, wherein the shape analysis comprises:

computing a compactness of the high intensity region; and

determining whether the high intensity region is a solid component or a vessel by using an area, a maximum distance on a distance transform map and the compactness of the region; or

determining whether the high intensity region is a solid component or a vessel by scaling and normalizing the region and computing a compactness for the scaled and normalized region.

2. The method of claim 1 , wherein performing an intensity-based segmentation comprises:

applying a high threshold to the segmented GGN to identify pulmonary structures having high intensity; and

after the high threshold has been applied, applying a low threshold to the segmented GGN to include boundary voxels that belong to the pulmonary structures having high intensity.

3. The method of claim 1 , wherein computing a compactness of the high intensity region comprises:

computing a compactness of the region in 2D; and

computing a compactness of the region in 3D.

4. The method of claim 1 , wherein determining whether the high intensity region is a solid component or a vessel by using an area, a maximum distance on a distance transform map and the compactness of the region comprises:

labeling the region as a vessel if its area is greater than a first area and the maximum distance is less than a first distance and the compactness is greater than a first compactness, or the maximum distance is less than a second distance and the compactness is greater than a second compactness, or the maximum distance is less than a third distance and the compactness is greater than a third compactness.

5. The method of claim 4 , further comprising:

setting a value of at least one of the first area, first through third distances and first through third compactnesses.

6. The method of claim 1 , wherein the compactness of the scaled and normalized region identifies the scaled and normalized region as a solid component or a vessel.

7. The method of claim 1 , further comprising:

acquiring a pulmonary image including a non-segmented GGN by using a computed tomography (CT) technique; and

segmenting the non-segmented GGN.

8. A system for detecting solid components in ground glass nodules (GGNs) in medical images, comprising:

a memory device for storing a program; and

a processor in communication with the memory device, the processor operative with the program to:

perform an intensity-based segmentation on a segmented GGN to identify a high intensity region; and

perform a shape analysis to determine whether the high intensity region is a solid component or a vessel, wherein when performing the shape analysis the processor is further operative with the program to:

compute a compactness of the high intensity region; and

determine whether the high intensity region is a solid component or a vessel by using an area, a maximum distance on a distance transform map and the compactness of the region; or

determine whether the high intensity region is a solid component or a vessel by scaling and normalizing the region and computing a compactness for the scaled and normalized region.

9. The system of claim 8 , wherein when performing an intensity-based segmentation the processor is further operative with the program to:

apply a high threshold to the segmented GGN to identify pulmonary structures having high intensity; and

after the high threshold has been applied, apply a low threshold to the segmented GGN to include boundary voxels that belong to the pulmonary structures having high intensity.

10. The system of claim 8 , wherein when computing a compactness of the high intensity region the processor is further operative with the program to:

compute a compactness of the region in 2D; and

compute a compactness of the region in 3D.

11. The system of claim 8 , wherein when determining whether the high intensity region is a solid component or a vessel by using an area, a maximum distance on a distance transform map and the compactness of the region the processor is further operative with the program to:

label the region as a vessel if its area is greater than a first area and the maximum distance is less than a first distance and the compactness is greater than a first compactness, or the maximum distance is less than a second distance and the compactness is greater than a second compactness, or the maximum distance is less than a third distance and the compactness is greater than a third compactness.

12. The system of claim 11 , wherein the processor is further operative with the program to:

set a value of at least one of the first area, first through third distances and first through third compactnesses.

13. The system of claim 8 , wherein the compactness of the scaled and normalized region identifies the scaled and normalized region as a solid component or a vessel.

14. The system of claim 8 , wherein the processor is further operative with the program to:

acquire a pulmonary image including a non-segmented GGN from a computed tomography (CT) scanner; and

segment the non-segmented GCN.

15. A method for detecting solid components in ground glass nodules (GGNs) in medical images, comprising:

performing an intensity-based segmentation on a segmented volume of interest (VOI) to identify a high intensity region, wherein the VOI includes a GGN;

performing a shape analysis to determine whether the high intensity region is a solid component of the GGN or a vessel, wherein the shape analysis comprises:

computing a compactness of the high intensity region; and

determining whether the high intensity region is a solid component of the GGN or a vessel by using an area, a maximum distance on a distance transform nap and the compactness of the region; or

determining whether the high intensity region is a solid component of the GGN or a vessel by scaling and normalizing the region and computing, a compactness for the scaled and normalized region.

16. The method of claim 15 , wherein performing an intensity-based segmentation comprises:

applying a high threshold to the segmented VOI to identify pulmonary structures having high intensity; and

after the high threshold has been applied, applying a low threshold to the segmented VOI to include boundary voxels that belong to the pulmonary structures having high intensity.

17. The method of claim 15 , wherein determining whether the high intensity region is a solid component of the GGN or a vessel by using an area, a maximum distance on a distance transform map and the compactness of the region comprises:

labeling the region as a vessel if its area is greater than a first area and the maximum distance is less than a first distance and the compactness is greater than a first compactness, or the maximum distance is less than a second distance and the compactness is greater than a second compactness, or the maximum distance is less than a third distance and the compactness is greater than a third compactness.

18. The method of claim 15 , wherein the compactness of the scaled and normalized region identifies the scaled and normalized region as a solid component or a vessel.

19. A system for detecting solid components in ground glass nodules (GGNs) in medical images, comprising:

a memory device for storing a program; and

a processor in communication with the memory device, the processor operative with the program to:

perform an intensity-based segmentation on a segmented volume of interest (VOI) to identify a high intensity region, wherein the VOI includes a GGN;

perform a shape analysis to determine whether the high intensity region is a solid component of the GGN or a vessel, wherein when performing the shape analysis the processor is further operative with the program to;

compute a compactness of the high intensity region; and

determine whether the high intensity region is a solid component of the GGN or a vessel by using an area, a maximum distance on a distance transform map and the compactness of the region; or

determine whether the high intensity region is a solid component of the GGN or a vessel by sealing and normalizing the region and computing a compactness for the scaled and normalized region.

20. The system of claim 19 , wherein when performing an intensity-based segmentation the processor is further operative with the program to:

apply a high threshold to the segmented VOI to identify pulmonary structures having high intensity; and

after the high threshold has been applied, apply a low threshold to the segmented VOI to include boundary voxels that belong to the pulmonary structures having high intensity.

21. The system of claim 19 , wherein when determining whether the high intensity region is a solid component of the GGN or a vessel by using an area, a maximum distance on a distance transform map and the compactness of the region the processor is further operative with the program to:

label the region as a vessel if its area is greater than a first area and the maximum distance is less than a first distance and the compactness is greater than a first compactness, or the maximum distance is less than a second distance and the compactness is greater than a second compactness, or the maximum distance is less than a third distance and the compactness is greater than a third compactness.

22. The system of claim 19 , wherein the compactness of the scaled and normalized region identifies the scaled and normalized region as a solid component or a vessel.