Apparatus for detecting bubbles in coverslip adhesive

• US 5,566,249 A
• Filed: 09/20/1994
• Issued: 10/15/1996
• Est. Priority Date: 09/20/1994
• Status: Expired due to Term

• Current Assignee: Tripath Imaging Incorporated (Becton, Dickinson & Co)
• Original Assignee: NeoPath, Inc. (Becton, Dickinson & Co)
• Inventors: Kuan, Chih-Chau L., Rosenlof, Mikel D.
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Tadayon, Bijan
• Application Number:
  US08/309,077
• Time in Patent Office: 756 Days
• Field of Search: 382/257, 382/283, 382/133, 382/134, 382/128, 382/144, 382/145, 382/146, 382/147, 382/149, 382/169, 382/170, 382/172, 382/173, 382/190, 382/192, 382/195, 382/199, 382/203 , 382/204, 382/224, 382/256, 382/258, 382/259, 382/260, 382/263, 382/308, 382/293, 382/291, 382/286, 382/282, 382/264, 382/266, 382/268, 382/275, 382/274, 382/271 , 382/269, 382/270
• US Class Current: 382/257
• CPC Class Codes: G06V 20/69 Microscopic objects, e.g. biological cells or cellular parts

Abstract

The detection of bubbles in coverslip adhesive. Bubbles in coverslip adhesive may span large areas of the coverslip. An automated cytology system acquires 4 x gray scale images of the slide. Image conditioning is then performed. The conditioned gray scale image is binomially filtered providing a binomial filtered image. The gray scale image is subtracted from the binomial filtered image to produce a high pass enhanced image. An object mask detection is then performed on the high pass enhanced image. Objects are classified to identify bubble edgest. Multiple detections of bubble edges are pieced together to detect bubbles much larger than the area covered by a single image from the detector and the microscope lens. Bubbles are located even if low level image analysis shows gaps in the edges. The cytological system fills in significant gaps in the detected bubble edges. The result is then used for further processing by the automated cytology system.

Claims

1. A bubble mapper apparatus comprising:

(a) imaging means for capturing an image of a slide to obtain a representation of the image;

(b) means for detecting bubble edges from images to generate a bubble edge output;

(c) dilation means for dilating the bubble edge output to generate a dilated bubble edge map;

(d) means for connected component analysis on the dilated bubble edge map to produce a labeled bubble map;

(e) means for horizontal and vertical filling the labeled bubble map to produce a binary bubble map; and

(f) erosion means for eroding the binary bubble map to produce a final bubble map.

2. The apparatus of claim 1 wherein the imaging means further comprises an automated microscope.

3. The apparatus of claim 2 wherein the automated microscope further comprises a charge coupled device array.

4. The apparatus of claim 1 wherein the dilation means further comprises a 3×

3 dilation.

5. The apparatus of claim 1 wherein the means for eroding further comprises a 3×

3 erosion.

6. The apparatus of claim 1 wherein the means for detecting bubble edges further comprises a bubble edge detection feature classifier.

7. The apparatus of claim 6 wherein the bubble edge detection feature classifier further comprises a Fisher linear feature combination thresholder.

8. The apparatus of claim 1 wherein the dilation means further comprises a digital computer.

9. The apparatus of claim 1 wherein the means for connected component analysis further comprises a digital computer.

10. The apparatus of claim 1 wherein the means for horizontal and vertical filling further comprises a digital computer.

11. The apparatus of claim 1 herein the erosion means further comprises a digital computer.

12. The apparatus of claim 1 wherein the slide is selected from the group consisting of a biological specimen slide, a gynecological specimen slide, and a slide prepared with a Papanicolaou method.

25. The apparatus of claim 1 further comprising a means for eliminating the area defined by the bubble map from further processing.

13. A method of detecting bubble edges comprising the steps of:

(a) obtaining a gray scale image of a biological specimen at a predetermined magnification;

(b) image conditioning the gray scale image to produce a conditioned gray scale image; and

(c) segmenting the conditioned gray scale image into objects that are similar to bubble edges.

14. A method of image segmentation comprising the steps of:

(a) obtaining a gray scale image of a biological specimen at a predetermined magnification;

(b) image conditioning the gray scale image to generate a conditioned gray scale image; and

(c) segmenting the conditioned gray scale image into objects that are similar to bubble edges.

15. The method of claim 14 further comprises the step of classifying the object as a bubble edge or as a non bubble edge.

16. The method of claim 15 wherein the step of classifying the object further comprises the step of bubble edge detection feature classification.

17. A bubble mapping method comprising the steps of:

(a) imaging a slide having a coverslip to obtain a representation of a biological specimen under the coverslip;

(b) detecting bubble edges from images to generate a bubble edge output;

(c) performing a 3×

3 dilation on the bubble edge output to generate a dilated bubble edge map;

(d) performing a connected component analysis on the dilated bubble edge map to produce a labeled bubble map;

(e) performing a horizontal and vertical fill on the labeled bubble map to produce a binary bubble map; and

(f) performing a 3×

3 erosion on the binary bubble map to produce a final bubble map.

18. The method of claim 17 wherein the slide is selected from the group consisting of a biological specimen slide, a gynecological specimen, and a slide prepared with a Papanicolaou method.

26. The method of claim 17 further comprising the step of eliminating the area defined by the bubble map from further processing.

19. A method for detecting bubble edges in an image of a slide having a coverslip comprising the steps of:

(a) acquiring a plurality of images of the coverslip, wherein the plurality of images tiles a region under the coverslip;

(b) image conditioning the plurality of images to generate a conditioned gray scale image;

(c) bionomially filtering each conditioned gray scale image to provide a binomially filtered image;

(d) subtracting the conditioned gray scale image from the binomial filtered image to produce a high pass filtered image;

(e) performing object detection on the high pass filtered image to detect objects; and

(f) classifying the objects to identify bubble edges.

20. The method of claim 19 further comprising the steps of:

(a) piecing the bubble edges from the plurality of images together to form a map of bubbled edges over the region under the coverslip;

(b) separating the bubbles in the map from each other; and

(c) filling the outlines of the distinct bubbles to include the interiors of each bubble.

21. The method of claim 19 wherein each of the plurality of images is approximately 1.408 mm by 1.408 mm in size.

22. The method of claim 19 wherein the plurality of images comprise 4 x gray scale images.

23. A bubble mapping method comprising the steps of:

(a) imaging a slide having a coverslip to obtain a representation of a biological specimen under the coverslip;

(b) detecting bubble edges from the representation of a biological specimen under the coverslip to generate a bubble edge output;

(c) performing a 3×

3 dilation on the bubble edge output to generate a dilated bubble edge map;

(d) performing a connected component analysis on the dilated bubble edge map to produce a labeled bubble map;

(e) performing a horizontal and vertical fill on the labeled bubble map to produce a binary bubble map; and

(f) performing a 3×

3 erosion on the binary bubble map to produce a final bubble map.

24. The method of claim 23 wherein the slide is selected from the group consisting of a biological specimen slide, a gynecological specimen, and a slide prepared with a Papanicolaou method.