# US 2011/0234812 A1 · Method and a system for determining a target in a biological sample by image analysis

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2011/0234812 A1
- Document ID: US20110234812A1
- Status: Published application
- Kind: A1
- Application number: 13/058,385
- Filed: 2009-08-14
- Published: 2011-09-29
- Earliest claimed: 2008-08-15
- Assignees: Visiopharm A/S
- Inventors: Michael Grunkin; Niels Taekker Foged
- Web page: https://patentgrep.com/patent/US20110234812A1

## CPC

- G16B 45/00: ICT specially adapted for bioinformatics-related data visualisation, e.g. displaying of maps or networks
- G06V 20/695: {Preprocessing, e.g. image segmentation}
- G06F 17/18: for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)}
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

The present invention relates to methods for identifying and optionally quantifying stained cells in a cell or tissue sample and determining expression level of a target, as well as methods for determining staining intensity and/or staining quality of a stained sample, calibrating digital imaging apparatus using the methods, as well as systems and a computer readable mediums therefore. In particular the invention relates to a method for identifying stained targets in a sample and uses of said method, said method comprising providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target, and filtering the digital representation with at least three filters, each of said filters comprising filter constants), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation, segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and from said segmented representation identifying the stained targets in the sample.

## Claims

### Claim 1

A method for identifying one or more stained targets in a sample, said method comprising:
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target, and
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and
d) from said segmented representation identifying the stained targets in the sample.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the filter constant(s) are estimated polynomial coefficients.

### Claim 3 (depends on 2)

The method according to claim 2, wherein the polynomial coefficients are estimated from a second order polynomial, a fourth order polynomial, or a sixth order polynomial.

### Claim 4 (depends on 1)

The method according to claim 1, wherein at least three filter constants are applied to the pixels in the subset.

### Claim 5 (depends on 1)

The method according to claim 1, wherein the subset correspond to the entire digital representation.

### Claim 6 (depends on 1)

The method according to claim 1, wherein the filter constant(s) is(are) determined prior to analysing a plurality of representations and the same constant(s) are used for each digital representation.

### Claim 7 (depends on 1)

The method according to claim 1, wherein the filter constant(s) is(are) determined for each digital representation.

### Claim 8 (depends on 2)

The method according to claim 2, wherein a Hessian matrix is determined for the polynomial coefficient, and the Eigensolution is determined from said Hessian matrix.

### Claim 9 (depends on 2)

The method according to claim 2, wherein a least-squares estimator is used to determine the coefficient(s) of the polynomial, such as a weighted least-squares estimator.

### Claim 10 (depends on 1)

The method according to claim 1, wherein the sample is a tissue sample.

### Claim 11 (depends on 1)

The method according to claim 1, wherein the sample is a sample suspected of comprising cancer cells.

### Claim 12 (depends on 11)

The method according claim 11, wherein the cancer cells are breast cancer cells.

### Claim 13 (depends on 1)

The method of claim 1, wherein the stain is a chromophore or a fluorophore.

### Claim 14 (depends on 1)

The method according to claim 1, wherein the stain stains for a target associated with the cell membrane, such as a protein associated with the cell membrane.

### Claim 15 (depends on 14)

The method of claim 14, wherein the protein is selected from the HER-family, cytokeratines; and CD-antigens.

### Claim 16 (depends on 1)

The method according to claim 1, wherein the stain stains for a target associated with linear structures outside the cell, such as a protein lining vessels or bronchi.

### Claim 17 (depends on 14)

The method according to claim 14, wherein the Eigensolution is the first Eigen value.

### Claim 18 (depends on 1)

The method according to claim 1, wherein the stain stains for at least one blob-shaped target in, on or outside the cell(s).

### Claim 19 (depends on 18)

The method according to claim 18, wherein the Eigensolution is the second Eigen value.

### Claim 20 (depends on 1)

The method according claim 1, wherein the filtered representation is normalized with the intensity representation.

### Claim 21 (depends on 1)

The method according to claim 1, wherein the segmented representation is post-processed.

### Claim 22 (depends on 21)

The method according to claim 21, wherein the post-processing relates to elimination of stained cell membranes, extracellular linear structures or blobs having an area below or higher than a predefined area.

### Claim 23 (depends on 1)

The method according to claim 1, wherein a measure for connectivity is computed as area of cell membranes divided with number of connected cell membranes.

### Claim 24 (depends on 1)

The method according to claim 1, wherein the staining intensity in the segmented representation is calculated.

### Claim 25 (depends on 24)

The method according to claim 24, wherein the staining intensity is calculated as the average staining intensity, such as with respect to a red chromaticity value or a normalized brown value.

### Claim 26 (depends on 2)

The method according to claim 2, wherein the window size for the polynomial is fitted to be larger than the size of the subset.

### Claim 27

A method for determining the staining intensity of one or more stained targets, said method comprising:
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target,
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and
from said segmented representation identifying the stained targets in the sample, and calculating the staining intensity in the digital representation.

### Claim 28

A method for quantifying the one or more stained targets, said method comprising:
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target,
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and
from said segmented representation identifying the stained targets in the sample, and quantifying the staining target(s) in the digital representation.

### Claim 29

A method for determining an expression level of one or more targets in a sample, said method comprising
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target,
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and
from said segmented representation identifying the stained targets in the sample, and calculating the staining intensity in the segmented representation, thereby determining the expression level of the target.

### Claim 30

A method for providing a calibration curve of staining intensity for one or more stained targets, said method comprising
providing a plurality of digital representations of known samples wherein each sample has been stained with a known amount of stain, wherein the sample has been stained with a stain staining a target,
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and
from said segmented representation identifying the stained targets in the sample, and calculating the staining intensity in the segmented representation for each digital representation, thereby obtaining a calibration curve for the staining intensity.

### Claim 31

A method for calibrating a digital imaging apparatus, comprising
providing a plurality of digital representations of known samples wherein each sample has been stained with a known amount of stain, wherein the sample has been stained with a stain staining one or more targets,
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and
from said segmented representation identifying the stained cell targets in the sample, and calculating the staining intensity in the segmented representation for each digital representation,
comparing the calculated staining intensity with a calibration curve for the stain intensity, thereby obtaining a calibration measure, and
calibrating the digital imaging apparatus based on said calibration measure.

### Claim 32

A method for evaluating the staining quality of one or more stained targets, said method comprising:
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target,
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation, and
calculating the staining intensity in the segmented representation for each digital representation,
comparing the calculated staining intensity with a calibration curve for the stain intensity, and
evaluating the staining quality of the stained targets.

### Claim 33

(canceled)

### Claim 34

A method for classifying a sample, said method comprising:
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target, and
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation,
d) from said segmented representation identifying the stained targets in the sample, and
based on the identified stained targets classifying the sample in relation to two or more groups of samples.

### Claim 35 (depends on 34)

The method according to claim 34 wherein the classification is performed by statistically calculating the likelihood that the sample belongs to a specific group.

### Claim 36 (depends on 34)

The method according to claim 34, wherein a quantification of stained targets are performed, and the classification is based on the quantification.

### Claim 37

A computer readable medium comprising:
instructions for,
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target, and
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation,
d) from said segmented representation identifying the stained targets in the sample.

### Claim 38

An automated system comprising, in combination:
a database capable of including a plurality of digital representations of a plurality of biological specimens;
a software module for analyzing a plurality of pixels from a digital representation of a biological specimen; and
a control module comprising instructions for:
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target, and
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation,
d) from said segmented representation identifying the stained targets in the sample.

### Claim 39 (depends on 38)

The automated system according to claim 38, wherein said system comprises a slide loader, a barcode reader, a microscope and a stage.

### Claim 40

A software program comprising:
instructions for:
providing at least one digital representation of the sample wherein the sample has been stained with a stain staining a target, and
a) filtering the digital representation with at least three filters, each of said filters comprising filter constant(s), applying said filter constant(s) to each pixels in at least a subset of the digital representation and determining an eigensolution based on the filter output for each pixel in the subset, wherein said filters are capable of enhancing the stained targets obtaining a filtered representation,
b) optionally normalizing the filtered representation,
c) segmenting the digital representation based on information from the optionally normalized filtered representation into stained cell targets and background, optionally combining with the digital representation, thereby obtaining a segmented representation, wherein said segmented representation discriminates between stained targets and other structures in the digital representation,
d) from said segmented representation identifying the stained targets in the sample.
