System and method for detection of ground glass objects and nodules

• US 7,548,642 B2
• Filed: 10/21/2005
• Issued: 06/16/2009
• Est. Priority Date: 10/28/2004
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Odry, Benjamin L.
• Primary Examiner(s): Kassa; Yosef
• Application Number:
  US11/255,826
• Publication Number:
  US 20060093216A1
• Time in Patent Office: 1,334 Days
• Field of Search: 382/128, 382/131, 382/141, 382/274
• US Class Current: 382/141
• CPC Class Codes: G06T 2207/30061 Lung
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/64 of convexity or concavity

Abstract

A method for detecting ground glass objects in a digitized image comprises the steps of determining intensity statistics from a plurality of ground glass objects, determining an intensity map of the distance of each point in an image from a mean intensity of said ground glass objects wherein one or more candidate ground glass objects can be identified, determining a circularity value for each candidate ground glass object, and calculating one or more feature values wherein said candidate ground glass object is characterized.

Claims

1. A method for detecting ground glass objects in a digitized medical image comprising the steps of:

providing a digitized medical image comprising a plurality of intensities corresponding to a domain of points on an N-dimensional grid;

determining intensity statistics from a plurality of ground glass objects;

determining an intensity map of the distance of each point in said image from a mean intensity of said ground glass objects wherein one or more candidate ground glass objects can be identified;

determining a circularity value for each candidate ground glass object; and

calculating one or more feature values wherein said candidate ground glass object is characterized, wherein said circularity value and features values are adapted to distinguishing an actual ground glass object from a candidate ground glass object.

2. The method of claim 1, wherein determining an intensity map for each point comprises:

dividing a range of intensities into intervals;

assigning each point in said image to an intensity interval;

determining a weighted intensity for each point based on the intensities of its nearest neighbors; and

calculating a function of the difference of each point intensity and the mean ground glass object intensity, wherein a contiguous collection of points are identified as being part of a candidate ground glass object.

3. The method of claim 2, wherein said function is

4. The method of claim 2, wherein the weighted intensity for each point is calculated from the formula

(Sum(neighborhood point intensities)+w %(point intensity))/(9+w),

where w is a predetermined weighting factor.

5. The method of claim 1, wherein determining a circularity value comprises smoothing said candidate ground glass object based on mathematical morphology and computing a ratio of the surface area of the candidate object to the area of a circle of a same radius.

6. The method of claim 1, further comprising growing a region for each said candidate ground glass object and computing a gray-level co-occurrence matrix of each region.

7. The method of claim 6, wherein growing a region further comprises determining a seed point for each candidate object, determining a distance map of each point in the candidate ground glass object to the background, applying a threshold to the distance map wherein points whose distance map exceeds the threshold are excluded from said region, and dilating said region.

8. The method of claim 6, wherein the N-dimensional grid is a 2D rectangular lattice, and said co-occurrence matrix is calculated for point pairs at angular orientations of 0°

, 45°

, 90°

, and 135°

.

9. The method of claim 8, wherein said co-occurrence matrix frequencies are sets defined by:

10. The method of claim 9, wherein the distance d is two points apart.

11. The method of claim 1, wherein said features comprise the mean, variance, skewness, kurtosis, energy, entropy, homogeneity, and contrast.

12. A method for a computer processor to detect ground glass objects in a digitized medical image, said method comprising the steps of:

determining an intensity map of the distance of each pixel in the medical image from a mean intensity of said ground glass objects wherein one or more contiguous collections of points are identified as candidate ground glass objects;

determining a circularity value for each candidate ground glass object;

calculating one or more intensity level co-occurrence matrices for each candidate object, wherein each co-occurrence matrix is calculated for pixel pairs at a predetermined pixel pair separation distance and angular orientation; and

calculating one or more feature values from said co-occurrence matrices.

13. The method of claim 12, further comprising analyzing one or more digitized images to determine a mean ground glass object intensity value and a standard deviation.

14. The method of claim 12, wherein said digitized image comprises a plurality of intensities corresponding to a domain of points on an 2-dimensional rectangular grid, said predetermined angles are 0°

, 45°

, 90°

, and 135°

, and said pixel pair separation distance is 2 pixels.

15. A program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform the method steps for detecting ground glass objects in a digitized medical image, said method comprising the steps of:

providing a digitized medical image comprising a plurality of intensities corresponding to a domain of points on an N-dimensional grid;

determining intensity statistics from a plurality of ground glass objects;

determining an intensity map of the distance of each point in said image from a mean intensity of said ground glass objects wherein one or more candidate ground glass objects can be identified;

determining a circularity value for each candidate ground glass object; and

calculating one or more feature values wherein said candidate ground glass object is characterized.

16. The computer readable program storage device of claim 15, wherein determining an intensity map for each point comprises:

dividing a range of intensities into intervals;

assigning each point in said image to an intensity interval;

determining a weighted intensity for each point based on the intensities of its nearest neighbors; and

calculating a function of the difference of each point intensity and the mean ground glass object intensity, wherein a contiguous collection of points are identified as being part of a candidate ground glass object.

17. The computer readable program storage device of claim 16, wherein said function is

18. The computer readable program storage device of claim 16, wherein the weighted intensity for each point is calculated from the formula

(Sum(neighborhood point intensities)+w %(point intensity))/(9+w),

where w is a predetermined weighting factor.

19. The computer readable program storage device of claim 15, wherein determining a circularity value comprises smoothing said candidate ground glass object based on mathematical morphology, and computing a ratio of the surface area of the candidate object to the area of a circle of a same radius.

20. The computer readable program storage device of claim 15, the method further comprising growing a region for each said candidate ground glass object and computing a gray-level co-occurrence matrix of each region.

21. The computer readable program storage device of claim 20, wherein growing a region further comprises determining a seed point for each candidate object, determining a distance map of each point in the candidate ground glass object to the background, applying a threshold to the distance map wherein points whose distance map exceeds the threshold are excluded from said region, and dilating said region.

22. The computer readable program storage device of claim 20, wherein the N-dimensional grid is a 2D rectangular lattice, and said co-occurrence matrix is calculated for point pairs at angular orientations of 0°

, 45°

, 90°

, and 135°

.

23. The computer readable program storage device of claim 22, wherein said co-occurrence matrix frequencies are sets defined by:

24. The computer readable program storage device of claim 23, wherein the distance d is two points apart.

25. The computer readable program storage device of claim 15, wherein said features comprise the mean, variance, skewness, kurtosis, energy, entropy, homogeneity, and contrast.