Targeted edge detection method and apparatus for cytological image processing applications

• US 8,064,679 B2
• Filed: 12/11/2008
• Issued: 11/22/2011
• Est. Priority Date: 12/27/2007
• Status: Active Grant

• Current Assignee: Gen-Probe Incorporated (Hologic Incorporated)
• Original Assignee: Cytyc Corporation (Hologic Incorporated)
• Inventors: Griffin, Christopher
• Primary Examiner(s): BARLOW JR, JOHN E
• Application Number:
  US12/332,562
• Publication Number:
  US 20090169090A1
• Time in Patent Office: 1,076 Days
• Field of Search: None
• US Class Current: 382/133
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06V 20/69 Microscopic objects, e.g. biological cells or cellular parts

Abstract

Edges in cytological image data are identified by obtaining a digital image of a specimen and computing a gradient image from the obtained digital image. A scaling function is applied to the grayscale image to identify regions of interest (e.g., edges of cell nuclei) in the digital image. Edges of the regions of interest are then identified based on the product of the computed gradient image and the scaling image. The scaling function may be applied to each image frame and one or more scaling thresholds are established for each frame to selectively pass, suppress, or scale pixels based on their measured intensity values. The scaled image resulting from application of the scaling function is multiplied with the gradient image to produce a targeted gradient image that identifies the edges of the region of interest. The targeted gradient image isolates edges corresponding to particular cellular structures, while rejecting other edges within the image.

Claims

1. A method of identifying edges of cellular structures contained in cytological image data comprising:

obtaining an intensity image of a specimen;

computing a gradient image from the obtained intensity image;

smoothing the intensity image to form a smoothed intensity image;

applying a scaling function to the smoothed intensity image to generate a scaling image identifying regions of interest; and

forming the product of the scaling image and gradient image so as to identify edges of the cellular structures of interest.

2. The method of claim 1, wherein the scaling function is updated for each frame of the digital image.

3. The method of claim 1, wherein the scaling function transforms image intensity into scaled values, wherein one or more ranges of image intensities are suppressed and one or more ranges of image intensities are retained.

4. The method of claim 3, wherein one or more ranges of image intensities are scaled in a substantially linear fashion.

5. The method of claim 4, wherein the ranges of image intensities to be preserved, suppressed, and scaled are updated for each frame.

6. The method of claim 1, the operation of identifying edges of regions of interest from the computed gradient image comprises mapping a scaling image to a Sobel gradient image.

7. The method of claim 1, wherein the scaling function comprises identifying pixels having intensity values within one or more ranges.

8. The method of claim 7, wherein the range(s) defining the scaling function is determined by k-means clustering of the pixel intensity values, in which class means are used to determine the ranges in the scaling function.

9. The method of claim 7, wherein the range of pixels is preset.

10. A method of identifying target edges contained in cytological image data comprising:

obtaining a digital image of a specimen, the digital image comprising a plurality of frames, each frame comprising pixels having varying intensity values;

applying a scaling function to the pixels in each frame to obtain a scaled image, wherein scaling function comprises an operation of retaining pixel intensity, suppressing pixel intensity, or applying a substantially linear transformation of pixel intensity based at least in part on pixel intensity values obtained from the digital image; and

mapping the scaled image to a gradient image of the digital image, wherein the mapped image identifies target edges contained in the cytological image data.

11. The method of claim 10, wherein the scaling function comprises establishing a range of pixel intensity values, wherein pixels having intensity values below the range are retained, pixels having intensity values above the range are suppressed, and pixels having intensity values within the range are scaled using the substantially linear transformation.

12. The method of claim 10, wherein the scaling function comprises establishing a range of pixel intensity values, wherein pixels having intensity values below the range are suppressed, pixels having intensity values above the range are retained, and pixels having intensity values within the range are scaled using the substantially linear transformation.

13. The method of claim 10, wherein the gradient image comprises a Sobel gradient image.

14. The method of claim 11, wherein the scaling function comprises determining the range using class means computed through k-means clustering of the pixel intensity values.

15. The method of claim 12, wherein the scaling function comprises determining the range using class means computed through k-means clustering of the pixel intensity values.

16. The method of claim 11, wherein the range of pixels is preset.

17. The method of claim 12, wherein the range of pixels is preset.

18. A system for generating cytological image data that identifies a target edge of a cellular feature contained therein, comprising:

an imaging station configured to obtain an image of a cytological specimen, the image comprising a plurality of image frames; and

one or more image processors configured to

calculate a Gradient Image of an image frame,

generate a Scaling Image of the image frame by application of a Scaling Function to each pixel in the frame, and

generate a Targeted Gradient Image based at least in part on the Gradient Image and the Scaling Image, the Targeted Gradient Image comprising cytological image data including the target edge of the cellular feature.

19. The system of claim 18, wherein the edge of the cellular feature comprises the nucleus edge.

20. The system of claim 18, wherein the edge of the cellular feature comprises the cytoplasm edge.