Method of, and apparatus and computer software for, performing image processing

• US 8,041,090 B2
• Filed: 08/16/2006
• Issued: 10/18/2011
• Est. Priority Date: 09/10/2005
• Status: Active Grant

• Current Assignee: Global Life Sciences Solutions Operations UK Ltd. (Danaher Corporation)
• Original Assignee: GE Healthcare UK Limited (Danaher Corporation)
• Inventors: Bye, Iain, Kalusche, Geert, Alexandrov, Yuriy
• Primary Examiner(s): Koval; Melissa
• Assistant Examiner(s): CHAN, EMILY Y
• Application Number:
  US12/065,744
• Publication Number:
  US 20080219529A1
• Time in Patent Office: 1,889 Days
• Field of Search: 382128-133, 382/203, 382/206, 382/224, 382/286, 435/4, 702 19- 21
• US Class Current: 382/128
• CPC Class Codes: G01N 15/1433 using image recognition
• CPC Class Codes: G01N 2015/1477 Multiparameters
• CPC Class Codes: G06F 18/211 Selection of the most significant subset of features
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06V 10/771 Feature selection, e.g. selecting representative features from a multi-dimensional feature space
• CPC Class Codes: G06V 20/698 Matching; Classification

Abstract

A computer-implemented method of performing image processing for images of biological objects includes: storing definitions of a plurality of descriptors; receiving image data relating to an image of a reference population of biological objects; receiving image data relating to an image of a target population of biological objects; processing the reference population image data to obtain a reference set of measurements, containing data for each of the descriptors; processing the target population image data to obtain a target set of measurements, containing data for each of the descriptors; and selecting a combination of the descriptors on the basis of comparing the reference set with the target set to define a preferred combination of the descriptors for use in identifying characteristics of a further population of biological objects which are similar to those of the target population.

Claims

1. A computer-implemented method of performing image processing for images of biological objects, said method comprising:

a) storing definitions of each of a plurality of descriptors for which measurements can be taken in relation to said biological objects;

b) receiving image data relating to an image of a reference population of biological objects exposed to a given biological environment;

c) receiving image data relating to an image of a target population of biological objects exposed to a different given biological environment;

d) processing said reference population image data to obtain a reference set of measurements, containing data for each of said plurality of descriptors;

e) processing said target population image data to obtain a target set of measurements, containing data for each of said plurality of descriptors; and

f) selecting a combination of one or more of said plurality of descriptors on the basis of comparing said reference measurement data with said target measurements data to define a preferred combination of said descriptors for use in identifying characteristics of a further population of biological objects which are similar to those of the target population.

2. The method of claim 1, further comprising:

g) modifying said sets of measurements, by changing the selected combination of descriptors,

h) comparing respective ones of said plurality of reference sets of measurements with corresponding ones of said plurality of target sets of measurements to determine a plurality of measures of difference between said reference population and said target population; and

i) comparing said measures of difference determined in step h) to select said preferred combination of said descriptors.

3. The method of claim 1, wherein a measure of difference between said reference population and said target population is maximised for a set of measurements of said reference population with said preferred combination of descriptors.

4. The method of claim 1, wherein at least one of said measurements relates to an intensity of said image of said reference population or to an intensity of said image of said target population.

5. The method of claim 1, wherein at least one of said measurements relates to a morphological characteristic of said objects in said image of said reference population or to a morphological characteristic of said objects of said target population.

6. The method of claim 1, wherein selecting the combination in step f) includes excluding at least one of said plurality of descriptors from said combination selected in step f), using said reference population measurements and said target population measurements, so that said preferred combination excludes at least one descriptor.

7. The method of claim 1, wherein said target population has been treated with a treatment agent and said reference population has not been treated with said treatment agent.

8. The method of claim 7, wherein said reference population is a control population of biological objects.

9. The method of claim 1, wherein each said biological object is a biological cell.

10. The method of claim 1, further comprising:

j) receiving further image data relating to an image of a further population of biological objects exposed to a given environment;

k) processing said further population image data to derive a set of measurements with said preferred combination of said descriptors; and

l) processing said set of measurements for said further population to identify characteristics which are similar to those of said target population.

11. The method of claim 10, wherein said processing in step l) includes classifying said objects of said further population.

12. The method of claim 11, wherein said processing in step l) includes identifying a cluster of said objects of said further population for said classification.

13. The method of claim 1, wherein a user provides input which is used to define said preferred combination of said descriptors.