Microcalcification detection classification in radiographic images

• US 8,855,388 B2
• Filed: 04/29/2011
• Issued: 10/07/2014
• Est. Priority Date: 04/30/2010
• Status: Active Grant

• Current Assignee: iCAD Incorporated
• Original Assignee: VuCOMP, Inc.
• Inventors: Wehnes, Jeffrey C., Monaco, James P., Harding, David S., Pike, James H., Ho, Anbinh T., Hanafy, Lawrence M.
• Primary Examiner(s): SHAH, UTPAL D
• Application Number:
  US13/695,347
• Publication Number:
  US 20130051676A1
• Time in Patent Office: 1,257 Days
• Field of Search: 382/128, 382/190, 382/205
• US Class Current: 382/128
• CPC Class Codes: G06F 18/24 Classification techniques
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 2207/30096 Tumor; Lesion
• CPC Class Codes: G06T 5/10 using non-spatial domain filtering
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06V 10/454 Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]

Abstract

An analysis of a digitized image is provided. The digitized image is repeatedly convolved to form first convolved images, which first convolved images are convolved a second time to form second convolved images. Each first convolved image and the respective second convolved image representing a stage, and each stage represents a different scale or size of anomaly. As an example, the first convolution may utilize a Gaussian convolver, and the second convolution may utilize a Laplacian convolver, but other convolvers may be used. The second convolved image from a current stage and the first convolved image from a previous stage are used with a neighborhood median determined from the second convolved image from the current stage by a peak detector to detect peaks, or possible anomalies for that particular scale.

Claims

1. A method for detecting an anomaly in an image, the method comprising:

convolving a digital image with a blurring filter to create a plurality of first convolved images at differing scales;

convolving each of the plurality of first convolved images with a curvature detection filter, thereby creating a plurality of second convolved images, each of the plurality of first convolved images and a corresponding one of the plurality of second convolved images corresponding to respective ones of a plurality of stages;

creating a plurality of surgery masks, each surgery mask being based at least in part on one of the plurality of the second convolved images of a current stage and one of the plurality of second convolved images from a previous stage;

determining a neighborhood median for each pixel location of the plurality of surgery masks; and

identifying one or more peaks in the digital image based at least upon the second convolved image from the current stage, the first convolved image from the previous stage, and the neighborhood medians for the current stage.

2. The method of claim 1, wherein the creating the plurality of surgery masks based at least in part on one of the plurality of second convolved images from the previous stage is performed at least in part by using the neighborhood median from the previous stage.

3. The method of claim 2, wherein the creating the plurality of surgery masks based at least in part on one of the plurality of second convolved images from the previous stage is performed at least in part by comparing, for each pixel of respective ones of the plurality of second convolved images, a pixel relative contrast to a threshold.

4. The method of claim 3, wherein the pixel relative contrast is determined at least in part by dividing a pixel value of the respective second convolved images of the current stage by a neighborhood median of a same pixel location of a previous stage.

5. The method of claim 1, further comprising excluding a peak detected at a smaller scale from a larger scale.

6. The method of claim 1, wherein the digital image is a scaled image.

7. The method of claim 1, wherein a standard deviation doubles from one scale to a next scale.

8. A system for identifying anomalies in a digitized image, the system comprising:

a processor; and

a non-transitory computer readable storage medium storing programming for execution by the processor, the programming including instructions to;

convolve a digitized image with a blurring filter, creating a first convolved image;

convolve the first convolved image with a curvature detection filter, creating a second convolved image;

determine a neighborhood median for each pixel in the second convolved image; and

identify pixels in the digitized image having peak values, in accordance with the second convolved image and the neighborhood median for each pixel in the second convolved image.

9. The system of claim 8, wherein the digitized image is a prior convolved image from a previous stage.

10. The system of claim 8, wherein the programming further includes instructions to remove pixels having a pixel relative contrast greater than a threshold from the second convolved image, and wherein the instructions to determine the neighborhood median for each pixel include further instructions to use at least in part the second convolved image after execution of the instructions to remove the pixels.

11. The system of claim 10, wherein the pixel relative contrast is determined at least in part by dividing a pixel value of the respective second convolved image of a current stage by a neighborhood median of a same pixel location based upon a previous stage second convolved image.

12. The system of claim 8, wherein the programming further includes instructions to compare corresponding peak values for the pixels to a threshold.

13. The system of claim 12, wherein the threshold is a multiple of a global median absolute deviation.

20. The system of claim 12, wherein the programming further includes instructions to generate a list of local maxima locations in the digitized image for the pixels having peak values higher than the threshold.

14. A computer program product for identifying anomalies, the computer program product having a non-transitory computer-readable medium with a computer program embodied thereon, the computer program comprising:

computer program code for convolving a digitized image with a blurring filter, thereby creating a plurality of first convolved images, each of the convolved images corresponding to a different scale;

computer program code for convolving each of the plurality of first convolved images with a curvature detection filter, creating a plurality of second convolved images;

computer program code for determining a neighborhood median for each pixel of each of the plurality of second convolved images; and

computer program code for identifying peak regions based upon one of the plurality of first convolved images and one of the second convolved images, the one of the second convolved images corresponding to the one of the plurality of first convolved images after being convolved by the convolving a digitized image and the convolving each of the plurality of first convolved images.

15. The computer program product of claim 14, wherein the plurality of first convolved images represent Gaussian-blurred images.

16. The computer program product of claim 14, wherein the plurality of second convolved images represent Laplacian curvature images.

17. The computer program product of claim 14, wherein the computer program code for determining the neighborhood median includes computer program code for excluding peak regions identified in a previous scale.

18. The computer program product of claim 14, further comprising computer program code for excluding pixels from the plurality of second convolved images used for determining the neighborhood median.

19. The computer program product of claim 18, wherein the computer program code for excluding pixels includes computer program code for comparing a pixel relative contrast to the neighborhood median.