Morphological classification system and method

• US 5,257,182 A
• Filed: 01/29/1991
• Issued: 10/26/1993
• Est. Priority Date: 01/29/1991
• Status: Expired due to Term

• Current Assignee: Tripath Imaging Incorporated (Becton, Dickinson & Co)
• Original Assignee: Neuromedical Systems, Inc.
• Inventors: Scott, Richard, Luck, Randall L.
• Primary Examiner(s): Envall, Jr., Roy N.
• Assistant Examiner(s): Brutman, Laura
• Application Number:
  US07/647,438
• Time in Patent Office: 1,001 Days
• Field of Search: 364/413.1, 364/413.08, 382/36, 395/924
• US Class Current: 382/224
• CPC Class Codes: G01N 15/1468 with spatial resolution of the texture or inner structure of the particle
• CPC Class Codes: G01N 15/1475 using image analysis for extracting features of the particle
• CPC Class Codes: G06V 20/693 Acquisition
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: Y10S 706/924 Medical

Abstract

A method of classifying cells based upon their morphology includes a method of classifying objects in a cytological specimen, including the steps of obtaining a first image of at least part of a cytological specimen, classifying objects in the first image on the basis of a predetermined criteria, obtaining a second image of at least one of the objects most likely to have a predetermined criteria, and displaying at least part of the second image to produce a visual display of at least one of the objects most likely to have a predetermined criteria. The method of classifying objects in a specimen includes the steps of obtaining a first digital representation of at least part of the cytological specimen, storing the first digital representation, performing a first filtering operation to filter out images in the first representation that are the approximate size of a malignant or premalignant cell or smaller to produce a second digital representation, removing the images in the second representation from the images in the first representation to produce a third representation, performing a second filtering operation to filter out images in the first representation that are the approximate size of premalignant or malignant cell or larger than the approximate size of a premalignant or malignant cell to produce a fourth representation, and eliminating the images in the fourth representation from the images in the third representation to produce a representation having substantially only images the approximate size of a premalignant or malignant cell.

Claims

1. A method of classifying objects in a cytological specimen, comprising the steps of:

a) obtaining a first image of at least part of such cytological specimen;

b) classifying objects in such first image on the basis of a predetermined criteria;

c) selecting at least one object for display based on said classifying;

d) obtaining a second image of at least part of such cytological specimen containing said at least one selected object; and

e) displaying at least part of such second image to produce a visual display of said at least one selected object.

2. The method of claim 1, wherein such first image is of a lower resolution than such second image.

3. The method of claim 1, including the step of further classifying such objects in such visual display.

4. The method of claim 1, wherein such visual display represents plural objects.

5. The method of claim 4, including the step of arranging such plural objects in such visual display in an eight by eight array of sixty-four elements.

6. The method of claim 5, wherein a centroid of each object displayed is approximately centered in one of such elements.

7. The method of claim 5, wherein such plural objects are arranged in an order related to the likelihood that each object has such predetermined criteria.

8. The method of claim 5, wherein such plural objects are arranged according to their relative positional locations in such cytological specimen.

9. The method of claim 4, including the step of arranging such plural objects in such visual display in a four by four array of sixteen elements.

10. The method of claim 9, wherein a centroid of each object displayed is approximately centered in one of such elements.

12. The method of claim 1 1, wherein said step of performing a first filtering operation includes performing a morphological well function.

13. The method of claim 1 1, wherein said step of performing a first filtering operation, includes performing a morphological top hat function.

16. The method of claim 1 1 wherein said step of performing a second filtering operation includes performing a morphological well function on such first digital representation.

17. The method of claim 1 1 wherein said step of performing a second filtering operation includes performing a morphological top hat function on such first digital representation.

18. The method of claim 1 1, wherein said step of removing includes masking out images which are too bright to be malignant or premalignant cells.

11. A method of classifying objects in a cytological specimen;

comprising the steps of;

a) obtaining a first digital representation of at least part of such cytological specimen;

b) storing such first digital representation;

c) performing a first filtering operation to filter out images in such representation that are the approximate size of a malignant or premalignant cell or smaller to produce a second digital representation;

d) removing the images in such second representation from the images in such first representation to produce a third representation;

e) performing a second filtering operation to filter out images in such first representation that are the approximate size of a premalignant or malignant cell or larger than the approximate size of premalignant or malignant cell to produce a fourth representation; and

f) eliminating the images in such fourth representation from the images in such third representation to produce a representation having substantially only images the approximate size of a premalignant or malignant cell.

14. The method of claim I 1 wherein said step of performing a second filtering operation includes performing an arithmetic gradient operation on such first digital representation.

15. The method of claim 14, wherein said step of performing an arithmetic gradient operation includes performing a Sobel operation.

19. The method of claim I 1, wherein said step of eliminating includes taking the complement of said fourth representation and boolean ANDing the resulting representation with such third representation.

20. The method of claim 19, wherein said step of taking the complement includes performing a boolean NOT.

21. An apparatus for classifying objects in a cytological specimen, comprising:

means for obtaining a first image of at least part of such cytological specimen;

means for classifying objects in such first image on the basis of a predetermined criteria;

means for selecting at least one object for display based on classification performed by said means for classifying;

means for obtaining a second image of such cytological specimen containing said at least one selected object; and

means for displaying at least part of such second image to produce a visual display of said at least one selected object.