Processing medical images of the breast to detect anatomical abnormalities therein
View Patent ↗Methods, systems and related computer program products are provided for processing a medical image of a breast to detect anatomical abnormalities therein, including anatomical abnormalities that may be associated with breast cancer. The medical image of the breast, which includes a background region bordering a breast tissue region along a skinline thereof, is processed to detect an inward-facing retraction along the skinline, which can be potentially indicative of an anatomical abnormality in the breast tissue. In one preferred embodiment, a display monitor displays first information representative of the medical image of the breast and second information identifying a location of the detected inward-facing retraction on the medical image of the breast. In another preferred embodiment, one or more metrics characterizing the detected inward-facing retraction are used as features in the classification of potential CAD detections in the breast tissue region.
1. A computer-implemented method for facilitating detection of anatomical abnormalities in a breast, comprising the computer-implemented steps of:
receiving a medical image of the breast, the medical image including a background region bordering a breast tissue region along a skinline thereof;
processing the medical image to detect an inward-facing retraction along said skinline of the breast tissue region and that satisfies pre-determined criteria, the inward-facing retraction being potentially indicative of an anatomical abnormality within the breast; and
displaying on a display monitor first information representative of said medical image of the breast and second information identifying a location of said detected inward-facing retraction on the medical image of the breast.
2. The computer-implemented method of claim 1 , wherein said second information comprises a graphical overlay that visually highlights said location of said inward-facing retraction on the medical image of the breast.
3. The computer-implemented method of claim 1 , wherein said processing the medical image to detect said inward-facing retraction comprises locating a contiguous segment of the skinline meeting predetermined criteria known to be at least partially associated with at least one of a shortening of milk ducts in the breast tissue and a shortening of glandular tissue strands in the breast tissue.
4. The computer-implemented method of claim 1 , wherein said processing the medical image to detect said inward-facing retraction comprises locating a contiguous segment of the skinline having a net average inward concavity greater than a predetermined threshold when evaluated over at least one of a plurality of predetermined averaging intervals along the skinline.
5. The computer-implemented method of claim 4 , wherein said predetermined threshold is between about 3 mm-10 mm and wherein at least one of said plurality of predetermined averaging intervals is between about 1 cm-5 cm.
6. The computer-implemented method of claim 4 , the pre-determined criteria comprising a threshold separation between contour lines generated based on respective points along said skinline, the method further comprising:
generating a relatively finely constrained contour line based on a plurality of relatively finely spaced anchor points along said skinline;
generating a relatively coarsely constrained contour line based on a plurality of relatively coarsely spaced anchor points along said skinline;
computing a separation between said relatively finely constrained contour line and said relatively coarsely constrained contour line as a function of distance therealong; and
identifying said inward facing retraction as a location along said skinline at which said separation is greater than a predetermined threshold separation over a predetermined threshold distance along said skinline.
7. The computer-implemented method of claim 1 , wherein said medical image of the breast is an x-ray mammogram.
8. The computer-implemented method of claim 1 , wherein said medical image of the breast is a two-dimensional image selected from the group consisting of: an x-ray mammogram, an x-ray tomosynthesis projection image, an x-ray tomosynthesis reconstructed image, a magnetic resonance imaging (MRI) slice, an MRI projection image, an x-ray computed tomography (CT) slice, an x-ray CT projection image, an ultrasound slice, and an ultrasound projection image.
9. The computer-implemented method of claim 7 , wherein said medical image results from an imaging procedure in which the breast is compressed along a compression plane, and wherein said medical image is representative of one of (i) a projection image acquired in a projection plane substantially parallel to said compression plane, and (ii) a slice of a three-dimensional image volume substantially parallel to said compression plane.
10. The computer-implemented method of claim 1 , further comprising:
processing the medical image to identify at least one candidate CAD detection in the breast tissue region;
computing at least one skinline retraction feature value for each said candidate CAD detection based at least in part on a spatial relationship between said candidate CAD detection and said detected inward-facing retraction; and
classifying each said candidate CAD detection according to a respective plurality of computed feature values including said at least one computed skinline retraction feature value.
11. A non-transitory computer readable medium embodied with a computer program product that when executed directs a computing apparatus to facilitate detection of anatomical abnormalities in a breast, comprising: computer code for receiving a medical image of the breast, the medical image including a background region bordering a breast tissue region along a skinline thereof; and computer code for processing the medical image to detect an inward-facing retraction along said skinline of the breast tissue region and that satisfies pre-determined criteria, the inward-facing retraction being potentially indicative of an anatomical abnormality within the breast.
12. The computer readable medium of claim 11 , further comprising computer code for displaying on a display monitor first information representative of said medical image of the breast and second information identifying a location of said detected inward-facing retraction on the medical image of the breast.
13. The computer readable medium of claim 11 , further comprising: computer code for processing the medical image to identify at least one candidate CAD detection in the breast tissue region; computer code for computing at least one skinline retraction feature value for each said candidate CAD detection based at least in part on a spatial relationship between said candidate CAD detection and said detected inward-facing retraction; and computer code for classifying each said candidate CAD detection according to a respective plurality of computed feature values including said at least one computed skinline retraction feature value.
14. The computer readable medium of claim 11 , wherein said computer code for processing the medical image to detect the inward-facing retraction comprises computer code for locating a contiguous segment of the skinline meeting predetermined criteria known to be at least partially associated with at least one of a shortening of milk ducts in the breast tissue and a shortening of glandular tissue strands in the breast tissue.
15. The computer readable medium of claim 11 , wherein said medical image of the breast is a two-dimensional image selected from the group consisting of: an x-ray mammogram, an x-ray tomosynthesis projection image, an x-ray tomosynthesis reconstructed image, a magnetic resonance imaging (MRI) slice, an MRI projection image, an x-ray computed tomography (CT) slice, an x-ray CT projection image, an ultrasound slice, and an ultrasound projection image.
16. A computer-aided detection (CAD) system for facilitating detection of anatomical abnormalities in a breast, comprising:
a first processor; and
a first memory associated with said first processor and storing first instructions that, when executed by said first processor, cause the first processor to perform steps comprising:
accessing a medical image of the breast, the medical image including a background region bordering a breast tissue region along a skinline thereof; and
processing the medical image to detect an inward-facing retraction along said skinline of the breast tissue region and that satisfies pre-determined criteria, the inward-facing retraction being potentially indicative of an anatomical abnormality within the breast.
17. The CAD system of claim 16 , further comprising:
a second processor; and
a second memory associated with said second processor and storing second instructions that, when executed by said second processor, cause the second processor to display on a display monitor first information representative of said medical image of the breast and second information identifying a location of said detected inward-facing retraction on the medical image of the breast.
18. The CAD system of claim 16 , said first instructions causing the first processor to perform further steps comprising:
processing the medical image to identify at least one candidate CAD detection in the breast tissue region;
computing at least one skinline retraction feature value for each said candidate CAD detection based at least in part on a spatial relationship between said candidate CAD detection and said detected inward-facing retraction; and
classifying each said candidate CAD detection according to a respective plurality of computed feature values including said at least one computed skinline retraction feature value.
19. The CAD system of claim 16 , wherein said processing the medical image to detect the inward-facing retraction comprises locating a contiguous segment of the skinline meeting predetermined criteria known to be at least partially associated with at least one of a shortening of milk ducts in the breast tissue and a shortening of glandular tissue strands in the breast tissue.
20. The CAD system of claim 16 , wherein said medical image of the breast is a two-dimensional image selected from the group consisting of: an x-ray mammogram, an x-ray tomosynthesis projection image, an x-ray tomosynthesis reconstructed image, a magnetic resonance imaging (MRI) slice, an MRI projection image, an x-ray computed tomography (CT) slice, an x-ray CT projection image, an ultrasound slice, and an ultrasound projection image.