Compartment segregation by pixel characterization using image data clustering

• US 8,655,037 B2
• Filed: 12/14/2012
• Issued: 02/18/2014
• Est. Priority Date: 05/14/2007
• Status: Active Grant

• Current Assignee: Novartis Ag
• Original Assignee: HistoRx, Inc. (NeoGenomics, Inc.)
• Inventors: Christiansen, Jason H., Pinard, Robert, Gustavson, Mark, Bourke, Brian, Reilly, Dylan M., Tedeschi, Gregory R.
• Primary Examiner(s): Laballe, Clayton E
• Assistant Examiner(s): Rhodes, Jr., Leon W
• Application Number:
  US13/715,937
• Publication Number:
  US 20130216111A1
• Time in Patent Office: 431 Days
• Field of Search: 382/128
• US Class Current: 382/128
• CPC Class Codes: G01N 33/5082 Supracellular entities, e.g. tissue, organisms
• CPC Class Codes: G01N 33/57423 of lung
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06V 20/69 Microscopic objects, e.g. biological cells or cellular parts

Abstract

The present invention relates generally to improved methods of defining areas or compartments within which biomarker expression is detected and quantified. In particular, the present invention relates to automated methods for delineating marker-defined compartments objectively with minimal operator intervention or decision making. The method provides for precise definition of tissue, cellular or subcellular compartments particularly in histological tissue sections in which to quantitatively analyzing protein expression.

Claims

1. A method for defining a first marker-defined biological compartment relative to a second marker-defined biological compartment present in a biological sample of interest, comprising

comparing the intensity in each of the pixel locations in a first image of the first marker defined biological compartment with the intensity in each of the corresponding pixel locations of a second image of the second marker-defined biological compartment, wherein the first image was prepared using a first imaging agent that is specific for the first marker-defined biological compartment, and wherein the second image was prepared using a second imaging agent that is specific for the second marker-defined biological compartment, wherein differences in pixel intensity define the first marker-defined biological compartment relative to the second marker-defined biological compartment,

assigning pixels to a cluster characterized by high first imaging agent intensity and low second imaging agent intensity to the first compartment;

assigning pixels to a cluster characterized by high second imaging agent intensity and low first imaging agent intensity to the second compartment; and

assigning pixels to a cluster characterized by low first imaging agent intensity and low second imaging agent intensity to background and removing such pixels from further analysis,

wherein any of the assigning steps are performed using a k means clustering method to determine a cluster membership for each pixel, and

wherein the method is automated.

2. The method of claim 1, wherein the method is implemented by a computer.

3. The method of claim 1, wherein the pixels of the two images are plotted, wherein the axes of the plot comprise the intensity of the first imaging agent and the intensity of the second imaging agent.

4. The method of claim 1, further comprising assigning remaining pixels with first imaging agent intensity and second imaging agent intensity to either the first compartment or the second compartment based on probability.

5. The method of claim 1, further comprising assigning those remaining pixels with first imaging agent intensity and second imaging agent intensity to neither the first compartment nor the second compartment.

6. The method of claim 1, wherein the biological compartment is selected from the group consisting of:

a cell type, sub-cellular compartment, a tissue compartment, and a localized cellular or tissue compartment.

7. The method of claim 6, wherein the biological compartment is a sub-cellular compartment selected from the group consisting of:

cell nucleus, cytoplasm, nuclear membrane, cellular membrane, mitochondria, endoplasmic reticulum, peroxisome and lysosome.

8. The method of claim 6, wherein the biological compartment is a tissue compartment selected from the group consisting of:

epithelium, stroma and mesothelia.

9. The method of claim 1, wherein the sample is a tissue sample, cell preparation or sub-cellular fraction.

10. The method of claim 1, further comprising defining a mask defined by the pixel intensity of the first and/or second imaging agent and defining compartment assignment for only those pixels within the mask.

11. The method of claim 1, further comprising incubating the sample with a first imaging agent that specifically labels the first marker-defined biological compartment, a second imaging agent that specifically labels a second marker-defined biological compartment.

12. A computer implemented method for defining a first marker-defined biological compartment relative to a second marker-defined biological compartment present in a biological sample comprising:

a) incubating the sample with a first imaging agent that specifically labels the first marker-defined compartment, and a second imaging agent that specifically labels the second marker defined compartment;

b) obtaining a first image of the first imaging agent labeled sample, and a second image of the second imaging agent labeled sample;

c) determining a first and a second imaging agent intensity in each corresponding pixel location in the first and the second image;

d) performing a clustering analysis on each pixel based on the first and the second imaging agent intensity of each pixel in each of the pixel to calculate clusters;

e) assigning those pixels in the cluster characterized by high first imaging agent pixel intensity and low second imaging agent pixel intensity to the first compartment;

f) assigning those pixels in the cluster characterized by high second imaging agent pixel intensity and low first imaging agent pixel intensity to the second compartment; and

g) assigning those pixels in the cluster characterized by low first imaging agent pixel intensity and low second imaging agent pixel intensity to background and removing such pixels from further analysis,

thereby defining a first marker-defined sub-cellular compartment relative to a second marker defined-sub-cellular compartment, wherein any of the assigning steps are performed using a k means clustering method to determine a cluster membership for each pixel, and

wherein the method is automated.

13. A computer implemented method for defining a first marker defined sub-cellular compartment relative to a second marker defined sub-cellular compartment present in individual cells of interest contained in a tissue sample comprising:

a) incubating the tissue sample with a first stain that specifically labels the first marker defined sub-cellular compartment, a second stain that specifically labels a second marker defined sub-cellular compartment,

b) obtaining an image of each of the first and the second stain in the tissue sample using a microscope so as to obtain;

i) a first image of the first marker defined sub-cellular compartment;

ii) a second image of the second marker defined sub-cellular compartment; and

c) determining the first and second stain intensity in each of the pixel locations in the first and the second image;

d) plotting the pixels, where the axes of the plot comprise the intensity of the first stain and the intensity of the second stain;

e) performing reiterative k-means clustering on the first and the second stain intensity in each of the pixel locations to calculate three clusters;

f) assigning those pixels in the cluster characterized by high first stain intensity and low second stain intensity to the first compartment;

g) assigning those pixels in the cluster characterized by high second stain intensity and low first stain intensity to the second compartment;

h) assigning those pixels in the cluster characterized by low first stain intensity and low second stain intensity to background and removing such pixels from further analysis;

i) assigning those pixels with first stain intensity and second stain intensity to either the first compartment or the second compartment based upon based on probability

thereby defining a first marker defined sub-cellular compartment relative to a second marker defined sub-cellular compartment, wherein any of the assigning steps are performed using a k means clustering method to determine a cluster membership for each pixel, and

wherein the method is automated.

14. A computer implemented method for defining a first marker defined sub-cellular compartment relative to a second marker defined sub-cellular compartment present in individual cells of interest contained in a tissue sample comprising:

a) incubating the tissue sample with a first stain that specifically labels the first marker defined sub-cellular compartment, a second stain that specifically labels a second marker defined sub-cellular compartment,

b) obtaining an image of each of the first and the second stain in the tissue sample using a microscope so as to obtain;

i) a first image of the first marker defined sub-cellular compartment;

ii) a second image of the second marker defined sub-cellular compartment; and

c) determining the first and second stain intensity in each of the pixel locations in the first and the second image;

d) plotting the pixels, where the axes of the plot comprise the intensity of the first stain and the intensity of the second stain;

e) performing reiterative k-means clustering on the first and the second stain intensity in each of the pixel locations to calculate three clusters;

f) assigning those pixels in the cluster characterized by high first stain intensity and low second stain intensity to the first compartment;

g) assigning those pixels in the cluster characterized by high second stain intensity and low first stain intensity to the second compartment;

h) assigning those pixels in the cluster characterized by low first stain intensity and low second stain intensity to background and removing such pixels from further analysis;

i) assigning those pixels with first stain intensity and second stain intensity to neither the first compartment or the second compartment;

thereby defining a first marker defined sub-cellular compartment relative to a second marker defined sub-cellular compartment, wherein any of the assigning steps are performed using a k means clustering method to determine a cluster membership for each pixel, and

wherein the method is automated.