Methods, Apparatuses, and Computer Program Products for Identifying a Region of Interest Within a Mammogram Image

• US 20140307937A1
• Filed: 06/27/2014
• Published: 10/16/2014
• Est. Priority Date: 07/13/2011
• Status: Active Grant

• Current Assignee: Change Healthcare Holdings, LLC (UnitedHealth Group Incorporated)
• Original Assignee: Mckesson Financial Holdings Limited (McKesson Corp.)
• Inventors: El-Hilo, Saba, Rezaee, Mahmoud Ramze
• Granted Patent: US 9,058,650 B2
• Time in Patent Office: Days
• US Class Current: 382/132
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/20156 Automatic seed setting
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding

Abstract

Methods, apparatuses, and computer program products are provided for identifying a region of interest within a mammogram image. A method may include applying a clustering algorithm to a histogram of the mammogram image to identify a predefined number of threshold values. The method may further include determining a predefined number of seed values based at least in part on the identified threshold values. The method may additionally include generating a kernel image for each of the seed values. The method may also include using the generated kernel images to identify a region of interest including a breast within the mammogram image. Corresponding apparatuses and computer program products are also provided.

Claims

1. A method for identifying a region of interest within a mammogram image, the method comprising:

applying a clustering algorithm to a histogram of the mammogram image to identify a predefined number of threshold values;

determining a predefined number of seed values based at least in part on the identified threshold values;

generating, by a processor, a kernel image for each of the seed values; and

using the generated kernel images to identify a region of interest comprising a breast within the mammogram image.

2. The method of claim 1, wherein applying the clustering algorithm to the histogram of the mammogram image comprises:

clustering the histogram into three clusters; and

identifying two threshold values on the basis of the three clusters.

3. The method of claim 2, wherein the first cluster represents a low intensity breast region including a breast skin line, the second cluster represents a medium intensity breast region including soft breast tissue, and the third cluster represents a high intensity breast region including any calcified regions of the breast.

4. The method of claim 1, wherein determining a predefined number of seed values comprises determining three seed values comprising a first seed value comprising a weighted mean representing low intensity breast region including a breast skin line, a second seed value comprising a weighted mean representing a medium intensity breast region including soft breast tissue, and a third seed value comprising a weighted mean representing a high intensity breast region including any calcified regions of the breast.

5. The method of claim 1, wherein generating a kernel image for each of the seed values comprises, for each seed value, applying a probability distribution kernel to the pixel values of the mammogram image.

6. The method of claim 1, wherein generating a kernel image for each of the seed values comprises:

generating a first kernel image having a distribution centered around a first seed value and representing pixels corresponding to a representing low intensity breast region including a breast skin line;

generating a second kernel image having a distribution centered around a second seed value and representing pixels corresponding to a medium intensity breast region including soft breast tissue; and

generating a third kernel image having a distribution centered around a third seed value and representing pixels corresponding to a high intensity breast region including any calcified regions of the breast.

7. The method of claim 1, wherein using the generated kernel images to identify a region of interest comprising a breast within the mammogram image comprises:

dividing the kernel images into equal sized regions;

fitting each region with a polynomial;

labeling each of a plurality of portions of the mammogram image as either breast or background based at least in part on a polynomial fit to a corresponding region of one or more of the kernel images to generate a binary image; and

identifying the region of interest comprising a breast within the mammogram image based at least in part on the binary image.

8. The method of claim 7, wherein labeling each of a plurality of portions of the mammogram image as either breast or background comprises labeling a portion breast in an instance in which a polynomial fit to a corresponding region of one or more of the kernel images has an order satisfying a threshold order.

9. The method of claim 7, wherein using the generated kernel images to identify a region of interest comprising a breast within the mammogram image further comprises:

determining a largest contiguous portion of the binary image labeled as breast; and

wherein identifying the region of interest comprising a breast within the mammogram image based at least in part on the binary image comprises identifying the region of interest comprising a breast within the mammogram image as the region within the mammogram image corresponding to the determined largest contiguous portion of the binary image labeled as breast.

10. An apparatus for identifying a region of interest within a mammogram image, the apparatus comprising at least one processor, wherein the at least one processor is configured to cause the apparatus to at least:

apply a clustering algorithm to a histogram of the mammogram image to identify a predefined number of threshold values;

determine a predefined number of seed values based at least in part on the identified threshold values;

generate a kernel image for each of the seed values; and

use the generated kernel images to identify a region of interest comprising a breast within the mammogram image.

11. The apparatus of claim 10, wherein the at least one processor is configured to cause the apparatus to apply the clustering algorithm to the histogram of the mammogram image at least in part by:

clustering the histogram into three clusters; and

identifying two threshold values on the basis of the three clusters.

12. The apparatus of claim 11, wherein the first cluster represents a low intensity breast region including a breast skin line, the second cluster represents a medium intensity breast region including soft breast tissue, and the third cluster represents a high intensity breast region including any calcified regions of the breast.

13. The apparatus of claim 10, wherein the at least one processor is configured to cause the apparatus to determine a predefined number of seed values at least in part by determining three seed values comprising a first seed value comprising a weighted mean representing low intensity breast region including a breast skin line, a second seed value comprising a weighted mean representing a medium intensity breast region including soft breast tissue, and a third seed value comprising a weighted mean representing a high intensity breast region including any calcified regions of the breast.

14. The apparatus of claim 10, wherein the at least one processor is configured to cause the apparatus to generate a kernel image for each of the seed values at least in part by, for each seed value, applying a probability distribution kernel to the pixel values of the mammogram image.

15. The apparatus of claim 10, wherein the at least one processor is configured to cause the apparatus to generate a kernel image for each of the seed values at least in part by:

generating a first kernel image having a distribution centered around a first seed value and representing pixels corresponding to a representing low intensity breast region including a breast skin line;

generating a second kernel image having a distribution centered around a second seed value and representing pixels corresponding to a medium intensity breast region including soft breast tissue; and

generating a third kernel image having a distribution centered around a third seed value and representing pixels corresponding to a high intensity breast region including any calcified regions of the breast.

16. The apparatus of claim 10, wherein the at least one processor is configured to cause the apparatus to use the generated kernel images to identify a region of interest comprising a breast within the mammogram image at least in part by:

dividing the kernel images into equal sized regions;

fitting each region with a polynomial;

labeling each of a plurality of portions of the mammogram image as either breast or background based at least in part on a polynomial fit to a corresponding region of one or more of the kernel images to generate a binary image; and

identifying the region of interest comprising a breast within the mammogram image based at least in part on the binary image.

17. The apparatus of claim 16, wherein labeling each of a plurality of portions of the mammogram image as either breast or background comprises labeling a portion breast in an instance in which a polynomial fit to a corresponding region of one or more of the kernel images has an order satisfying a threshold order.

18. The apparatus of claim 16, wherein the at least one processor is further configured to cause the apparatus to:

determine a largest contiguous portion of the binary image labeled as breast; and

wherein identifying the region of interest comprising a breast within the mammogram image based at least in part on the binary image comprises identifying the region of interest comprising a breast within the mammogram image as the region within the mammogram image corresponding to the determined largest contiguous portion of the binary image labeled as breast.

19. The apparatus of claim 10, further comprising at least one memory storing instructions that when executed by the at least one processor cause the apparatus to:

apply a clustering algorithm to a histogram of the mammogram image to identify a predefined number of threshold values;

determine a predefined number of seed values based at least in part on the identified threshold values;

generate a kernel image for each of the seed values; and

use the generated kernel images to identify a region of interest comprising a breast within the mammogram image.

20. A computer program product for identifying a region of interest within a mammogram image, the computer program product comprising at least one non-transitory computer-readable storage medium having computer-readable program instructions stored therein, the computer-readable program instructions comprising:

program instructions configured to apply a clustering algorithm to a histogram of the mammogram image to identify a predefined number of threshold values;

program instructions configured to determine a predefined number of seed values based at least in part on the identified threshold values;

program instructions configured to generate a kernel image for each of the seed values; and

program instructions configured to use the generated kernel images to identify a region of interest comprising a breast within the mammogram image.