Bronchial wall thickening recognition for reduced false-positives in pulmonary nodule detection

1. A method for automatically reducing false-positive nodule candidates associated with airways, the method comprising:

receiving a nodule candidate;

testing for airway cavities connected to said candidate; and

recognizing said candidate as a false-positive nodule candidate if it is connected to an airway cavity,

wherein said testing comprises:

perpendicular testing for airways that are relatively perpendicular to an examination plane; and

parallel testing for airways that are relatively parallel to an examination plane,

wherein said perpendicular testing comprises:

defining a region of interest comprising the nodule candidate;

thresholding the pixels within the region of interest;

connecting the thresholded pixels corresponding to a potential airway cavity; and

performing compactness and area tests on the potential airway cavity and classifying the candidate as true if an airway having sufficient compactness and area is not detected.

2. A method as defined in claim 1 wherein said examination plane is a viewing plane comprising one of a spin plane, a sagittal plane, or a coronal plane.

3. A method as defined in claim 1 wherein the threshold value for airway lumens is about −874 HU for computed tomography images.

4. A method as defined in claim 1 wherein said perpendicular testing further comprises:

performing an airway wall existence test for a potential airway cavity having sufficient compactness and area; and

classifying the candidate as true if the airway wall existence has a weak probability or false if the existence has a strong probability.

5. A method as defined in claim 4 wherein:

the probability is strong if the ratio of the number of airway wall pixels over the total number of outer boundary pixels is greater than about 80%; and

the probability is weak if the ratio of the number of non-wall pixels over the total number of outer boundary pixels is greater than about 50%.

6. A method as defined in claim 4 wherein said airway wall existence test comprises checking for pixels having gray levels between about 224 and 424 HU for computed tomography images.

7. A method as defined in claim 4 wherein said perpendicular testing further comprises:

performing an adjacent vessel test for a potential airway cavity having a medium probability of airway wall existence; and

classifying the candidate as true if there is no adjacent vessel detected or false if there is an adjacent vessel detected.

8. A method as defined in claim 7 wherein said adjacent vessel test comprises measuring distances between vessels and the potential airway cavity.

9. A method as defined in claim 7 wherein said adjacent vessel test comprises checking for adjacent vessels between about 2.5 mm and 4.5 mm from the potential airway cavity.

10. A method for automatically reducing false-positive nodule candidates associated with airways, the method comprising:

receiving a nodule candidate;

testing for airway cavities connected to said candidate; and

recognizing said candidate as a false-positive nodule candidate if it is connected to an airway cavity,

wherein said testing comprises;

perpendicular testing for airways that are relatively perpendicular to an examination plane; and

parallel testing for airways that are relatively parallel to an examination plane,

wherein said parallel testing comprises:

determining a profile direction by ridge detection;

assigning a confidence number for the middle pixel of each potential airway profile;

connecting pixels having high confidence numbers and deciding that the candidate is not located on an airway if the number of connected pixels is less than a threshold;

calculating the average confidence level of the connected pixels to decide that the candidate is not located on an airway if the average confidence level of the connected pixels is less than a threshold or that the candidate is located on an airway if the average confidence level of the connected pixels is not less than a threshold.

11. A method as defined in claim 10 wherein the threshold value for average confidence level is about 75%.

12. A method as defined in claim 10 wherein said testing further comprises:

testing all spin planes to detect an airway;

testing sagittal planes to detect an airway; and

testing coronal planes to detect an airway.

13. A system for automatically reducing false-positive nodule candidates associated with airways, the system comprising:

means for receiving a nodule candidate;

means for testing for airway cavities connected to said candidate, and

means for recognizing said candidate as a false-positive nodule candidate if it is connected to an airway cavity,

wherein said means for testing comprises:

means for perpendicular testing for airways that are relatively perpendicular to an examination plane; and

means for parallel testing for airways that are relatively parallel to an examination plane,

wherein said means for perpendicular testing comprises:

means for defining a region of interest comprising the nodule candidate;

means for thresholding the pixels within the region of interest;

means for connecting the thresholded pixels corresponding to a potential airway cavity; and

means for performing compactness and area tests on the potential airway cavity and means for classifying the candidate as true if an airway having sufficient compactness and area is not detected.

14. A system as defined in claim 13 wherein said means for perpendicular testing further comprises:

means for performing an airway wall existence test for a potential airway cavity having sufficient compactness and area; and

means for classifying the candidate as true if the existence has a weak probability or false if the existence has a strong probability.

15. A system as defined in claim 14 wherein said means for perpendicular testing further comprises:

means for performing an adjacent vessel test for a potential airway cavity having a medium probability of airway wall existence; and

means for classifying the candidate as true if there is no adjacent vessel detected or false if there is an adjacent vessel detected.

16. A system for automatically reducing false-positive nodule candidates associated with airways, the system comprising:

means for receiving a nodule candidate;

means for testing for airway cavities connected to said candidate; and

means for recognizing said candidate as a false-positive nodule candidate if it is connected to an airway cavity,

wherein said means for testing comprises:

means for perpendicular testing for airways that are relatively perpendicular to an examination plane; and

means for parallel testing for airways that are relatively parallel to an examination plane,

wherein said means for parallel testing comprises:

means for determining a profile direction by ridge detection;

means for assigning a confidence number for the middle pixel of each potential airway profile;

means for connecting pixels having high confidence numbers and deciding that the candidate is not located on an airway if the number of connected pixels is less than a threshold;

means for calculating the average confidence level of the connected pixels to decide that the candidate is not located on an airway if the average confidence level of the connected pixels is less than a threshold or that the candidate is located on an airway if the average confidence level of the connected pixels is not less than a threshold.

17. A system as defined in claim 16 wherein said means for testing further comprises:

means for testing all spin planes to detect an airway;

means for testing sagittal planes to detect an airway; and

means for testing coronal planes to detect an airway.

18. A computer readable medium encoded with a computer program embodying a program of instructions executable by a computer to perform method steps for automatically reducing false-positive nodule candidates associated with airways, the method steps comprising:

receiving a nodule candidate;

testing for airway cavities connected to said candidate; and

recognizing said candidate as a false-positive nodule candidate if it is connected to an airway cavity,

wherein said testing comprises:

perpendicular testing for airways that are relatively perpendicular to an examination plane; and

parallel testing for airways that are relatively parallel to an examination plane,

wherein said perpendicular testing comprises:

defining a region of interest comprising the nodule candidate;

thresholding the pixels within the region of interest;

connecting the thresholded pixels corresponding to a potential airway cavity; and

performing compactness and area tests on the potential airway cavity and classifying the candidate as true if an airway having sufficient compactness and area is not detected.

19. A computer readable medium as defined in claim 18 wherein said perpendicular testing further comprises:

performing an airway wall existence test for a potential airway cavity having sufficient compactness and area; and

classifying the candidate as true if the existence has a weak probability or false if the existence has a strong probability.

20. A computer readable medium as defined in claim 19 wherein said perpendicular testing further comprises:

performing an adjacent vessel test for a potential airway cavity having a medium probability of airway wall existence; and

classifying the candidate as true if there is no adjacent vessel detected or false if there is an adjacent vessel detected.

21. A computer readable medium encoded with a computer program embodying a program of instructions executable by a computer to perform method steps for automatically reducing false-positive nodule candidates associated with airways, the method steps comprising:

receiving a nodule candidate;

testing for airway cavities connected to said candidate; and

recognizing said candidate as a false-positive nodule candidate if it is connected to an airway cavity,

wherein said testing comprises:

perpendicular testing for airways that are relatively perpendicular to an examination plane; and

parallel testing for airways that are relatively parallel to an examination plane,

wherein said parallel testing comprises:

determining a profile direction by ridge detection;

assigning a confidence number for the middle pixel of each potential airway profile;

connecting pixels having high confidence numbers and deciding that the candidate is not located on an airway if the number of connected pixels is less than a threshold;

calculating the average confidence level of the connected pixels to decide that the candidate is not located on an airway if the average confidence level of the connected pixels is less than a threshold or that the candidate is located on an airway if the average confidence level of the connected pixels is not less than a threshold.

22. A computer readable medium as defined in claim 21 wherein said testing further comprises:

testing all spin planes to detect an airway;

testing sagittal planes to detect an airway; and

testing coronal planes to detect an airway.