Method for moving cell detection from temporal image sequence model estimation

• US 20080112606A1
• Filed: 11/09/2006
• Published: 05/15/2008
• Est. Priority Date: 11/09/2006
• Status: Active Grant

• Current Assignee: Leica Microsystems Gmbh (Danaher Corporation)
• Original Assignee: Shih-Jong J. Lee
• Inventors: Oh, Seho, Lee, Shih-Jong J.
• Granted Patent: US 8,045,783 B2
• Time in Patent Office: Days
• US Class Current: 382/133
• CPC Class Codes: G06T 2207/10056 Microscopic image
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: G06V 20/695 Preprocessing, e.g. image segmentation

Abstract

A computerized robust cell kinetic recognition method for moving cell detection from temporal image sequence receives an image sequence containing a current image. A dynamic spatial-temporal reference generation is performed to generate dynamic reference image output. A reference based object segmentation is performed to generate initial object segmentation output. An object matching and detection refinement is performed to generate kinetic recognition results output. The dynamic spatial-temporal reference generation step performs frame look ahead and the reference images contain a reference intensity image and at least one reference variation image.

Claims

1. A computerized robust cell kinetic recognition method for moving cell detection from temporal image sequence comprising the steps of:

a) Inputting an image sequence containing a current image;

b) Performing dynamic spatial-temporal reference generation using the image sequence having dynamic reference image output;

c) Performing reference based object segmentation having initial object segmentation output.

2. The robust cell kinetic recognition method of claim 1 further comprises a previous frame results storage and performing object matching and detection refinement using the initial object segmentation and the previous frame results having kinetic recognition results output.

3. The robust cell kinetic recognition method of claim 1 wherein the dynamic spatial-temporal reference generation step performing frame look ahead.

4. The robust cell kinetic recognition method of claim 1 wherein the reference based object segmentation method subtracting the dynamic reference image from the current image.

5. The robust cell kinetic recognition method of claim 1 wherein the dynamic spatial-temporal reference generation method comprising the steps of:

a) Inputting running interval image sequence;

b) Performing pixel statistics creation using the running interval image sequence having pixel statistics output;

c) Performing background time points detection using the running interval image sequence and the pixel statistics having background set output;

d) Performing reference image generation using the current image and the background set having reference intensity image output.

6. The dynamic spatial-temporal reference generation method of claim 5 further comprising the steps of:

a) Performing at least one spatial-temporal variation enhancement using the current image having at least one variation image output;

b) Performing reference image generation using the at least one variation image having at least one reference variation image output.

7. A computerized reference based object segmentation method for moving cell detection from temporal image sequence comprising the steps of:

a) Inputting an image sequence;

b) Inputting a reference intensity image and at least one reference variation image;

c) Performing differencing using the image sequence, the reference intensity image and the at least one reference variation image having a plurality of difference images output;

d) Performing multi-channel integration using the plurality of difference images having initial object segmentation output.

8. The reference based object segmentation method of claim 7 wherein the differencing step performing an operation selected from the set consisting of:

i) Absolute difference;

ii) Signed subtraction;

iii) One side clip subtraction;

iv) Run-off subtraction.

9. The reference based object segmentation method of claim 7 wherein the multi-channel integration step including adjustment for different sensitivity and specificity operating points.

10. The reference based object segmentation method of claim 7 wherein the multi-channel integration step performing an operation selected from the set consisting of:

i) Maximum operation followed by thresholding;

ii) Minimum operation followed by thresholding;

iii) Thresholding followed by Union;

iv) Thresholding followed by intersection.

11. A computerized object matching and detection refinement method for moving cell detection from temporal image sequence comprising the steps of:

a) Inputting an initial object segmentation;

b) Inputting previous frame results;

c) Performing matching and shift estimation using the initial object segmentation and the previous frame results having estimated shift and updated tracks or new cell detection output;

d) Performing aligned combination using the previous frame results and the estimated shift having combined previous results output.

12. The object matching and detection refinement method of claim 11 further comprising the steps of

a) Performing sub-region cutting using the initial object segmentation having cut object components output;

b) Performing integration using the cut object components and the combined previous results having integrated mask output.

13. The object matching and detection refinement method of claim 12 further performing final refinement using the integrated mask having final object segmentation output.

14. The object matching and detection refinement method of claim 11 wherein the matching and shift estimation step finding the match between objects of the initial object segmentation and objects of the previous frame results.

15. The matching and shift estimation step of claim 14 wherein the match is performed using at least one feature selected from the set consisting of location, static object morphology, intensity, object velocity and object acceleration having a goodness of match output.

16. The matching and shift estimation step of claim 14 further performing new cell detection.

17. The object matching and detection refinement method of claim 11 wherein the aligned combination performing a weighed combination wherein weights are goodness of matches of the previous frame results.

18. The object matching and detection refinement method of claim 12 wherein the integration step including a conflict resolution step.

19. The object matching and detection refinement method of claim 12 wherein the integration step including the cut object components close to the combined previous results.

20. The final refinement method of claim 13 performing structure guided processing.