Medical image review workstation with integrated content-based resource retrieval

• US 7,616,793 B2
• Filed: 12/30/2004
• Issued: 11/10/2009
• Est. Priority Date: 12/30/2004
• Status: Expired due to Fees

• Current Assignee: R2 Technology, Inc. (Hologic Incorporated)
• Original Assignee: Hologic Incorporated
• Inventors: Hartman, Keith W., Roehrig, Jimmy R., Marshall, Julian
• Primary Examiner(s): Strege; John B
• Application Number:
  US11/026,603
• Publication Number:
  US 20060147099A1
• Time in Patent Office: 1,776 Days
• Field of Search: 382/128, 382/130, 705/3, 707/6
• US Class Current: 382/128
• CPC Class Codes: G16H 30/20 for handling medical images, e.g. DICOM, HL7 or PACS
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 70/00 ICT specially adapted for the handling or processing of medical references
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

A review workstation for facilitating interpretation of a medical image by a user is described. The review workstation comprises a display device displaying the medical image to the user, and an input device receiving a graphical identification of a region of interest (ROI) in the medical image that is interesting to the user. Responsive to the identification of the ROI, a content-based image comparison is performed between the ROI and a resource database, the resource database comprising a collection of resource images previously analyzed by human interpreters and textual information associated with those previous analyses. The content-based image comparison comprises identifying a subset of the resource images similar to the ROI with respect to a preselected set of computed features. The display device subsequently displays to the user at least one of the identified subset of resource images and its associated textual information simultaneously with the ROI.

Claims

1. A workstation for facilitating interpretation of a medical image by a user, comprising:

a display device displaying the medical image to the user;

an input device receiving a graphical identification of a location of interest on the medical image from the user; and

a processor performing a content-based image comparison of the location of interest against a resource database, said resource database comprising a collection of resource images previously analyzed by human interpreters and textual information associated with said previous analyses, said content-based image comparison comprising identifying a subset of said resource images similar to said location of interest with respect to a preselected set of computed features;

wherein said resource database further comprises information categorizing each said resource image therein as being from one of;

(i) a first category of medical data source consisting exclusively of preselected, published medical reference texts and medical teaching materials authored by an identified core set of expert authors and commonly recognized as credible and authoritative; and

(ii) a second category of medical data source not qualified as being in said first category;

and wherein said display device displays to the user at least one of said identified subset of resource images and its associated textual information simultaneously with said location of interest, said display device further displaying information identifying into which of said first and second categories the displayed resource image is categorized.

2. The workstation of claim 1, wherein at least one of the resource images from said first category of medical data source is replaced in said resource database by an original, high-quality digital or digitized image corresponding thereto acquired through the author or a successor-in-interest thereof.

3. The workstation of claim 1, said processor further performing the steps of:

automatically generating textual keywords descriptive of said location of interest based on said preselected set of computed features;

automatically text-searching said textual information associated with said previously analyzed resource images based on said textual keywords; and

displaying to the user at least one of said collection of resource images whose associated textual information is identified by said automatic text search.

4. The workstation of claim 1, wherein said medical image is a mammogram, and wherein said preselected set of computed features includes one or more of size, spiculatedness, margin sharpness, eccentricity, sphericity, average grey level, contrast, cluster characteristics, and breast density characteristics.

5. The workstation of claim 4, wherein said preselected set of features may be altered by the user.

6. The workstation of claim 1, wherein an ordered list of selectable links corresponding to said identified subset of resource images is displayed to the user, the user selecting one of said selectable links to cause said simultaneous display of the corresponding resource image and the location of interest.

7. The workstation of claim 6, wherein said second category of medical data source includes at least one of (i) unpublished resource images previously analyzed by human interpreters and associated textual information as contained in a non-public archive, and (ii) web-posted resource images previously analyzed by human interpreters and associated textual information.

8. The workstation of claim 1, wherein said displayed resource images that are from said first category of medical data source and their associated textual information are displayed in their native published format.

9. The workstation of claim 1, said display device displaying computer-aided detection (CAD) markers on said medical image, said graphical identification of a location of interest comprising user selection of one of said CAD markers.

10. The workstation of claim 1, said display device displaying computer-aided detection (CAD) markers on said medical image, said graphical identification of a location of interest comprising one or more of (a) user selection of one of said CAD markers, (b) user selection of a point location, wherein said location of interest comprises a predetermined pattern including said point location, and (c) user circumscription of a region corresponding to the location of interest, whereby either (b) or (c) may result in an identified location of interest that does not include any of said CAD markers.

11. A method for facilitating interpretation of a medical image, comprising:

identifying a first plurality of medical data sources meeting a predetermined selection criterion as being from published medical reference texts and medical teaching materials authored by an identified core set of expert authors and commonly recognized as credible and authoritative;

identifying a second plurality of medical data sources not meeting said predetermined selection criterion;

forming from said first and second pluralities of medical data sources a resource database comprising a collection of resource images previously analyzed by human interpreters and textual information associated therewith, wherein said resource database further comprises information categorizing each of the resource images into one of a first category and a second category according to the one of the first and second pluralities of medical data sources, respectively, with which its corresponding medical data source was identified;

displaying the medical image to a user;

receiving a graphical input from the user identifying a region of interest (ROI) in the medical image that is interesting to the user;

performing a content-based image comparison of the ROI against said resource database, said content-based image comparison comprising identifying a subset of said resource images similar to said ROI with respect to a predetermined computed image feature set; and

displaying to the user at least one of said identified subset of resource images and its associated textual information simultaneously with said ROI, and further displaying information identifying into which of the first and second categories the displayed resource image is categorized.

12. The method of claim 11, wherein at least one of the resource images categorized in said first category is replaced in said resource database by an original, high-quality digital or digitized image corresponding thereto acquired through the author or a successor-in-interest thereof.

13. The method of claim 11, further comprising:

automatically generating textual keywords descriptive of said ROI based on said predetermined computed image feature set;

automatically text-searching said textual information associated with said previously analyzed resource images based on said textual keywords; and

displaying to the user at least one of said collection of resource images whose associated textual information is identified by said automatic text search.

14. The method of claim 11, wherein said medical image is a mammogram, and wherein said predetermined computed image feature set includes one or more of size, spiculatedness, margin sharpness, eccentricity, sphericity, average grey level, contrast, cluster characteristics, and breast density characteristics.

15. The method of claim 14, wherein said predetermined computed image feature set may be augmented, reduced, and/or edited by the user.

16. The method of claim 11, further comprising:

displaying an ordered list of selectable links corresponding to said identified subset of resource images, wherein said ordered list includes information identifying into which of the first and second categories each listed resource image is categorized;

receiving a user selection of one of said selectable links; and

displaying the resource image corresponding to the selected link and the textual information associated therewith simultaneously with said ROI.

17. The method of claim 16, wherein said resource images categorized in said second category are acquired from at least one of (i) unpublished resource images previously analyzed by human interpreters and associated textual information as contained in a non-public archive, and (ii) web-posted resource images previously analyzed by human interpreters and associated textual information.

18. The method of claim 11, wherein said displayed resource images categorized in said first category and their associated textual information are displayed in their native published format.

19. The method of claim 11, further comprising displaying computer-aided detection (CAD) markers on said medical image, said CAD markers being generated by a CAD algorithm identifying anatomical abnormalities in the medical image based at least in part upon said predetermined computed image feature set, wherein said receiving the graphical input comprises receiving a user selection of one of said CAD markers.

20. The method of claim 11, further comprising displaying computer-aided detection (CAD) markers on said medical image, said CAD markers being generated by a CAD algorithm identifying anatomical abnormalities in the medical image based at least in part upon said predetermined computed image feature set, wherein said receiving the graphical input comprises one or more of (a) receiving a user selection of one of said CAD markers, (b) receiving a user selection of a point location, wherein said ROI comprises a predetermined pattern including said point location, and (c) receiving a user circumscription of the ROI, whereby either (b) or (c) may result in an identified ROI that does not include any of said CAD markers.

21. A CAD system for facilitating interpretation of a medical image, comprising:

a CAD processor for processing the medical image to detect anatomical abnormalities therein according to a CAD algorithm, the CAD algorithm computing a predetermined feature set for each potentially abnormal location within the medical image;

a display unit for displaying the medical image to a user and further displaying thereon CAD markers corresponding to detected anatomical abnormalities;

a graphical input unit for receiving a graphical identification of a region of interest (ROI) in the medical image that is interesting to the user;

a comparison processor for performing a content-based image comparison of the ROI against a resource database, said resource database comprising a collection of resource images previously analyzed by human interpreters and textual information associated therewith, said content-based image comparison comprising identifying a subset of said resource images similar to said ROI with respect to one or more features in said predetermined feature set, said display unit displaying to the user at least one of said identified subset of resource images and said associated textual information simultaneously with said ROI;

wherein said resource database further comprises information categorizing each said resource image therein as being from one of;

(i) a first category of medical data source consisting exclusively of preselected, published medical reference texts and medical teaching materials authored by an identified core set of expert authors and commonly recognized as credible and authoritative; and

(ii) a second category of medical data source not qualified as being in said first category;

and wherein said display unit further displays information identifying into which of said first and second categories the displayed resource image is categorized.

22. The CAD system of claim 21, wherein at least one of the resource images categorized in said first category is replaced in said resource database by an original, high-quality digital or digitized image corresponding thereto acquired through the author or a successor-in-interest thereof.

23. The CAD system of claim 21, wherein said medical image is a mammogram, and wherein said predetermined feature set includes one or more of size, spiculatedness, margin sharpness, eccentricity, sphericity, average grey level, contrast, cluster characteristics, and breast density characteristics.

24. The CAD system of claim 23, wherein said predetermined feature set of may be altered by the user.

25. The CAD system of claim 21, wherein graphical input unit receives one or more of (a) a user selection of one of said CAD markers, (b) a user selection of a point location, wherein said ROI comprises a predetermined pattern including said point location, and (c) a user circumscription of the ROI, whereby either (b) or (c) may result in an identified ROI that does not include any of said CAD markers.

26. The CAD system of claim 21, wherein said resource images categorized in said second category are acquired from at least one of (i) unpublished resource images previously analyzed by human interpreters and associated textual information as contained in a non-public archive, and (ii) web-posted resource images previously analyzed by human interpreters and associated textual information.

27. The CAD system of claim 26, wherein, prior to said simultaneous display, said display unit displays an ordered list of selectable links corresponding to said identified subset of resource images, said ordered list including information identifying into which of the first and second categories each listed resource image is categorized, said graphical input unit receiving a user selection of one of said selectable links.

28. The CAD system of claim 27, wherein said resource database is editable by the user and/or a user institution to add resource images/text, delete resource images/text, and/or recategorize a resource image and its associated text into a different category.

29. The CAD system of claim 21, wherein said displayed resource image and its associated textual information is displayed in its native published format if said displayed resource image is categorized in said first category.