System and Method for Detection of Breast Masses and Calcifications Using the Tomosynthesis Projection and Reconstructed Images

• US 20080025592A1
• Filed: 06/25/2007
• Published: 01/31/2008
• Est. Priority Date: 06/27/2006
• Status: Active Grant

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Zhou, Xiang, Krishnan, Arun, Jerebko, Anna
• Granted Patent: US 7,840,046 B2
• Time in Patent Office: Days
• US Class Current: 382/132
• CPC Class Codes: A61B 6/03 Computed tomography [CT] echo-tomography A61B8/14
• CPC Class Codes: A61B 6/466 adapted to display 3D data
• CPC Class Codes: A61B 6/502 for diagnosis of breast, i.e. mammography
• CPC Class Codes: G06T 11/005 Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
• CPC Class Codes: G06T 2207/10112 Digital tomosynthesis [DTS]
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 2211/436 Limited angle
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/97 Determining parameters from multiple pictures depth or shape recovery from multiple images G06T7/55; stereo camera calibration G06T7/85

Abstract

A method of detecting breast masses and calcifications in digitized images, includes providing a plurality of 2-dimensional (2D) digital X-ray projectional breast images acquired from different viewing angles, extracting candidate lesions and 2D features from said 2D projectional images, computing spicularity characteristics of said candidate lesions, including location, periodicity, and amplitude, applying learning algorithms to said candidate lesions to predict a probability of malignancy of said lesion, receiving from said learning algorithm a probability map of detections for each breast image, said detections comprising associating pixels with a probability of being associated with a malignancy, creating a synthetic 2D slice for each X-ray image wherein malignant regions are indicated by ellipses on a non-malignant background, and constructing a synthetic 3-dimensional (3D) image volume from said 2D synthetic slices.

Claims

1. A method of detecting breast masses and calcifications in digitized images, comprising the steps of:

providing a plurality of 2-dimensional (2D) digital X-ray projectional breast images acquired from different viewing angles, each said X-ray projectional image comprising a set of intensities on a 2D grid of pixels;

extracting candidate lesions and 2D features from said 2D projectional images;

computing spicularity characteristics of said candidate lesions, said characteristics including, but not limited to, location, periodicity, and amplitude;

applying machine learning algorithms to said candidate lesions to predict a probability of malignancy of said lesion, wherein said learning algorithms have been trained on a training set containing known examples of the features for lesions identified by a physician;

receiving from said learning algorithm a probability map of detections for each breast image, said detections comprising associating pixels with a probability of being associated with a malignancy;

creating a synthetic 2D slice for each X-ray image wherein malignant regions are indicated by ellipses on a non-malignant background; and

constructing a synthetic 3-dimensional (3D) image volume from said 2D synthetic slices.

2. The method of claim 1, wherein said probability map associates each pixel on each 2D X-ray image with a probability of being associated with a malignancy.

3. The method of claim 1, wherein said probability map returns a list of pixel locations and associated malignancy probabilities, wherein only those locations with malignancy probabilities above a threshold are returned.

4. The method of claim 1, further comprising constructing a 3D image from said 2D X-ray breast images, and superimposing said 3D synthetic image on said constructed 3D X-ray image.

5. The method of claim 1, further comprising thresholding said detections in said 3D synthetic image wherein only those detections whose probability is above said threshold are visualized.

6. The method of claim 1, wherein the malignancy probability in said malignant regions are represented by Gaussian functions inside said ellipses, wherein a peak amplitude of each said Gaussian corresponds to the pixel within said ellipse with a maximum probability of being associated with said malignancy, and whose probability amplitude decreases smoothly to a background value.

7. The method of claim 6, wherein constructing said synthetic 3D image comprises seeking those 2D ellipses in those 2D synthetic slices that overlap in said 3D synthetic image, representing those 3D regions formed by overlapping 2D ellipses by 3D ellipsoids, representing the malignancy probability within each 3D ellipsoid by a 3D Gaussian whose probability amplitude decreases smoothly from a peak value within said ellipsoid to a background value.

8. The method of claim 1, wherein said spicularity characteristics are computed using a spherical or polar coordinate transformation on said lesion followed by band pass filtering on the intensity or gradient of said transformed lesion.

9. The method of claim 1, further comprising enhancing said X-ray breast images prior to extracting said 2D features.

10. A method of detecting breast masses and calcifications in digitized images, comprising the steps of:

providing a plurality of 2-dimensional (2D) digital X-ray breast images acquired from different viewing angles, said X-ray image comprising a set of intensities on a 2D grid of pixels;

constructing a 3-dimensional (3D) image volume from said 2D X-ray breast images;

computing spicularity characteristics of said candidate lesions, said characteristics including, but not limited to, location, periodicity, and amplitude;

extracting 3D features from said image volume using a machine learning algorithm;

applying machine learning algorithms to said candidate lesions to predict a probability of malignancy of said lesion, wherein said learning algorithms have been trained on a training set containing known examples of the features for lesions identified by a physician;

receiving from said learning algorithm a probability map of detections for each breast image, said detections comprising associating pixels with a probability of being associated with a malignancy; and

thresholding said detections in said 3D image volume wherein only those detections whose probability is above said threshold are visualized.

11. The method of claim 10, further comprising extracting 2D features from 2D image volume slices using machine learning algorithms, and providing these 2D features as inputs to said CAD algorithm.

12. The method of claim 10, further comprising enhancing said X-ray breast images prior to constructing said 3D image volume.

13. The method of claim 10, further comprising enhancing said 3D image volume prior to extracting 3D features from said image volume.

14. 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 breast masses and calcifications in digitized images, said method comprising the steps of:

providing a plurality of 2-dimensional (2D) digital X-ray projectional breast images acquired from different viewing angles, each said X-ray projectional image comprising a set of intensities on a 2D grid of pixels;

extracting candidate lesions and 2D features from said 2D projectional images;

computing spicularity characteristics of said candidate lesions, said characteristics including, but not limited to, location, periodicity, and amplitude;

applying machine learning algorithms to said candidate lesions to predict a probability of malignancy of said lesion, wherein said learning algorithms have been trained on a training set containing known examples of the features for lesions identified by a physician;

receiving from said learning algorithm a probability map of detections for each breast image, said detections comprising associating pixels with a probability of being associated with a malignancy;

creating a synthetic 2D slice for each X-ray image wherein malignant regions are indicated by ellipses on a non-malignant background; and

constructing a synthetic 3-dimensional (3D) image volume from said 2D synthetic slices.

15. The computer readable program storage device of claim 14, wherein said probability map associates each pixel on each 2D X-ray image with a probability of being associated with a malignancy.

16. The computer readable program storage device of claim 14, wherein the probability map returns a list of pixel locations and associated malignancy probabilities, wherein only those locations with malignancy probabilities above a threshold are returned.

17. The computer readable program storage device of claim 14, the method further comprising constructing a 3D image from said 2D X-ray breast images, and superimposing said 3D synthetic image on said constructed 3D X-ray image.

18. The computer readable program storage device of claim 14, the method further comprising thresholding said detections in said 3D synthetic image wherein only those detections whose probability is above said threshold are visualized.

19. The computer readable program storage device of claim 14, wherein the malignancy probability in said malignant regions are represented by Gaussian functions inside said ellipses, wherein a peak amplitude of each said Gaussian corresponds to the pixel within said ellipse with a maximum probability of being associated with said malignancy, and whose probability amplitude decreases smoothly to a background value.

20. The computer readable program storage device of claim 19, wherein constructing said synthetic 3D image comprises seeking those 2D ellipses in those 2D synthetic slices that overlap in said 3D synthetic image, representing those 3D regions formed by overlapping 2D ellipses by 3D ellipsoids, representing the malignancy probability within each 3D ellipsoid by a 3D Gaussian whose probability amplitude decreases smoothly from a peak value within said ellipsoid to a background value.

21. The computer readable program storage device of claim 14, wherein said spicularity characteristics are computed using a spherical or polar coordinate transformation on said lesion followed by band pass filtering on the intensity or gradient of said transformed lesion.

22. The computer readable program storage device of claim 14, the method further comprising enhancing said X-ray breast images prior to extracting said 2D features.