Breast cancer screening with adjunctive ultrasound mammography

• US 7,556,602 B2
• Filed: 05/31/2002
• Issued: 07/07/2009
• Est. Priority Date: 11/24/2000
• Status: Active Grant

• Current Assignee: U-System Inc. (General Electric Company)
• Original Assignee: U-System Inc. (General Electric Company)
• Inventors: Wang, Shih-Ping, Rao, Fangyi, Chin, Donald
• Primary Examiner(s): Le; Long V
• Assistant Examiner(s): Cattungal; Sanjay
• Application Number:
  US10/160,836
• Publication Number:
  US 20030007598A1
• Time in Patent Office: 2,594 Days
• Field of Search: 600337-461, 382/128, 382/132, 382/284, 382/294, 378/37, 378/63
• US Class Current: 600/437
• CPC Class Codes: A61B 6/4423 related to hygiene or sterilisation
• CPC Class Codes: A61B 6/463 characterised by displaying multiple images or images and diagnostic data on one display
• CPC Class Codes: A61B 6/502 for diagnosis of breast, i.e. mammography
• CPC Class Codes: A61B 6/5247 combining images from an ionising-radiation diagnostic technique and a non-ionising radiation diagnostic technique, e.g. X-ray and ultrasound
• CPC Class Codes: A61B 6/563 involving image data transmission via a network
• CPC Class Codes: A61B 8/0825 for diagnosis of the breast, e.g. mammography
• CPC Class Codes: A61B 8/406 using means for diagnosing suspended breasts
• CPC Class Codes: A61B 8/462 characterised by constructional features of the display
• CPC Class Codes: A61B 8/463 characterised by displaying multiple images or images and diagnostic data on one display composite display in devices for short-range imaging with acoustic waves G01S7/52074
• CPC Class Codes: A61B 8/464 involving a plurality of displays
• CPC Class Codes: A61B 8/465 adapted to display user selection data, e.g. icons or menus
• CPC Class Codes: A61B 8/467 characterised by special input means
• CPC Class Codes: A61B 8/469 for selection of a region of interest
• CPC Class Codes: A61B 8/5238 for combining image data of patient, e.g. merging several images from different acquisition modes into one image for image registration algorithms G06T7/30
• CPC Class Codes: A61B 8/565 involving data transmission via a network

Abstract

Ultrasound mammography in which an automated transducer scans the patient'"'"'s breast to generate many images of thin slices that are processed into few images of thick slices that can be displayed simultaneously for practical rapid assessment of the breast. The thin-slice images can be acquired by a technician so that a physician need only spend time in assessing the few displayed thick-slice images and, possibly, only a few of the thin-slice images that might match a suspected anomaly indicated in the thick-slice images. Computer-aided detection or diagnosis (CAD) can be performed on the images and resulting mark and/or other information can be displayed as well. Vibration images can be obtained as well and similarly processed and displayed to highlight abnormalities or for other reasons.

Claims

1. A method of ultrasound examination of a patient'"'"'s breast, comprising:

compressing the patient'"'"'s breast along a compression direction by exerting a compressive force;

acquiring image data representing a plurality of thin-slice ultrasound images by scanning the patient'"'"'s breast in an automated scanning process in a scanning direction with an ultrasound transducer that is long in a direction transverse to the scanning direction, wherein;

said thin-slice images conform to respective planes that are generally along the compression direction; and

each of said thin-slice images represents an image of a thin slice of the patient'"'"'s breast;

computer-processing the image data representing said thin-slice images into image data representing a plurality of thick-slice images that are fewer in number than said thin slices, wherein;

said thick-slice images conform to respective planes that are transverse to said compression direction; and

each of said thick-slice images represents a thick slice of the patient'"'"'s breast that is thicker than a slice represented by one of said thin-slice images; and

displaying a plurality of said thick-slice images on a computerized display to thereby present for viewing medical information regarding the patient'"'"'s breast in fewer displayed images than if said thin-slice images were displayed, and where each displayed image contains more information than a single thin-slice image.

2. The method of claim 1 wherein each of said thick-slice images represents a slice of the breast having a thickness between about 2 mm and 20 mm.

3. The method of claim 2, wherein said thick-slice images represents slices of the breast that collectively occupy a contiguous majority of a volume of the breast.

4. The method of claim 2, wherein said thick-slice images represents slices of the breast that collectively occupy a screening volume of the breast, said screening volume consisting of the entire breast volume minus those portions known to have statistically insignificant occurrences of cancerous lesions.

5. The method of claim 4, said portions known to have statistically insignificant occurrences of cancerous lesions including all breast locations within 0.5 inches of a skin surface of the breast.

6. The method of claim 2, further comprising:

performing computer-assisted diagnosis (CAD) algorithms for each of said thick-slice images; and

overlaying markers on said thick-slice images on said first computerized display at locations corresponding to suspected lesions identified by said CAD algorithms.

7. The method of claim 6, said performing CAD algorithms comprising:

locating two-dimensional regions of interest (ROIs) in said thick-slice images according to a two-dimensional ROI location algorithm;

segmenting two-dimensional borders of candidate lesions at each of said two-dimensional ROIs;

extracting two-dimensional x-ray CAD features for each of said candidate lesions; and

classifying said candidate lesions based on said two-dimensional x-ray CAD features.

8. The method of claim 7, said two-dimensional x-ray CAD features being selected from the group consisting of:

spiculation metrics, density/contrast metrics, eccentricity metrics, circularity metrics, border roughness metrics, location metrics, and tenting metrics.

9. The method of claim 8, said performing CAD algorithms further comprising:

extracting two-dimensional acoustical CAD features for each of said candidate lesions; and

classifying said candidate lesions based on (i) said two-dimensional x-ray CAD features, and (ii) said two-dimensional acoustical CAD features.

10. The method of claim 9, said two-dimensional acoustical CAD features being selected from the group consisting of:

lateral shadow metrics, vertical shadow metrics, and posterior enhancement metrics.

11. The method of claim 10, said performing CAD algorithms using information from both (i) a given one of said thick-slice images, and (ii) a thick-slice volume represented by at least two adjacent thin-slice images the image data of which was computer-processed into the image data representing said given thick-slice image,

locating three-dimensional ROIs in said thick-slice volume according to a three-dimensional ROI location algorithm;

segmenting three-dimensional borders of candidate lesions at each of said three-dimensional ROIs;

extracting three-dimensional acoustical CAD features for each of said candidate lesions;

registering each three-dimensional ROI with zero or more of said two-dimensional ROIs; and

classifying, for any three-dimensional ROIs registering with one or more two-dimensional ROIs, the corresponding candidate lesions based on (i) said two-dimensional x-ray CAD features, and (ii) said two-dimensional acoustical CAD features, and (iii) said three-dimensional acoustical CAD features.

12. The method of claim 11, said three-dimensional acoustical CAD features being selected from the group consisting of:

surface roughness metrics, surface area-to-volume ratios, lesion compression metrics, volumetric echo uniformity metrics;

three-dimensional spiculation metrics, three-dimensional density metrics, sphericity metrics, and shadow metrics.

13. The method of claim 11, said registering said three-dimensional ROIs with said two-dimensional ROIs comprising:

computing a nipple distance metric for each of said three-dimensional ROIs and said two-dimensional ROIs; and

positively registering any of said three-dimensional ROIs with any of said two-dimensional ROIs if their associated nipple distance metrics differ by less than a first predetermined amount.

14. The method of claim 11, said registering said three-dimensional ROIs with said two-dimensional ROIs comprising:

computing a nipple distance metric for each of said three-dimensional ROIs and said two-dimensional ROIs;

computing a nipple angle metric for each of said three-dimensional ROIs and said two-dimensional ROIs; and

positively registering any of said three-dimensional ROIs with any of said two-dimensional ROIs if (i) their associated nipple distance metrics differ by less than a first predetermined amount; and

(ii) their associated nipple distance metrics differ by less than a second predetermined amount.

15. The method of claim 11, said registering said three-dimensional ROIs with said two-dimensional ROIs comprising:

computing a nipple distance metric for each of said three-dimensional ROIs and said two-dimensional ROIs;

computing a chest wall distance metric for each of said three-dimensional ROIs and said two-dimensional ROIs; and

positively registering any of said three-dimensional ROIs with any of said two-dimensional ROIs if (i) their associated nipple distance metrics differ by less than a first predetermined amount; and

(ii) their associated chest wall distance metrics differ by less than a third predetermined amount.

16. The method of claim 6, further comprising:

locating, segmenting, and extracting two-dimensional features of one or more two-dimensional regions of interest (ROIs) from said thick-slice images;

acquiring a second set of image data representing ultrasound Doppler frames acquired while the breast is being vibrated at one or more audio frequencies;

generating one or more vibrational Doppler images for each of said thick-slice images based on said second set of image data;

extracting a vibrational Doppler feature from said one or more vibrational Doppler images corresponding to each of said two-dimensional ROIs in said thick-slice images; and

classifying each of said two-dimensional ROIs based on said two-dimensional features and said vibrational Doppler feature.

17. The method of claim 16, wherein said vibrational Doppler feature comprises a vibrational resonance feature.

18. The method of claim 16, said two-dimensional features comprising two-dimensional x-ray CAD features and two-dimensional acoustical CAD features.

19. The method of claim 6, further comprising:

locating, segmenting, and extracting acoustic features of one or more regions of interest (ROIs) from each thick-slice image;

receiving a second set of image data representing ultrasound Doppler frames acquired while the breast is being vibrated at one or more audio frequencies;

generating a Vibrational Doppler volume for each of said thick-slice images based on said second set of image data plurality;

extracting a vibrational Doppler feature from said vibrational Doppler volume corresponding to each ROI in said thick-slice volume; and

classifying each ROI based on said acoustic features and said vibrational Doppler feature.

20. The method of claim 1, further comprising:

receiving a second set of image data representing ultrasound Doppler frames acquired while the breast is being vibrated at one or more audio frequencies;

generating a vibrational Doppler overlay image for each of said thick-slice images based on said second set of image data;

receiving a vibrational Doppler overlay command from a user; and

overlaying said vibrational Doppler overlay images onto said thick-slice images responsive to said vibrational Doppler overlay command.

21. The method of claim 20, wherein said vibrational Doppler overlays comprise vibrational resonance information.

22. The method of claim 1, said computer-processing including an integration comprising an arithmetic mean of image data of all individual ultrasound thin-slices within said thick-slice.

23. The method of claim 1, said computer-processing including integration comprising a statistical combination of image data of all thin-slice images within each of said thick-slices, said statistical combination being selected from the group consisting of:

weighted and unweighted arithmetic mean, weighted and unweighted geometric mean, weighted and unweighted reciprocal mean, weighted and unweighted exponential mean, maximum value, minimum value, standard deviation, and variance.

24. The method of claim 2, wherein said thick-slices have identical thicknesses.

25. The method of claim 2, wherein said thick-slices conform to planes parallel to a standard x-ray mammogram view plane.