Method and system for classifying and identifying individual cells in a microscopy image

Patent metadata

Field	Value
Publication number	US-9830502-B2
Application number	US-201415031523-A
Country	US
Kind code	B2
Filing date	Oct 14, 2014
Priority date	Oct 28, 2013
Publication date	Nov 28, 2017
Grant date	Nov 28, 2017

Abstract

In a method and system for identifying objects in an image, an image and training data are received. The training data identifies a pixel associated with an object of a particular type in the image. A plurality of filtered versions of the image are developed. The training data and the plurality of filtered versions of the image are processed to develop a trained model for classifying pixels associated with objects of the particular type. The trained model is applied to the image to identify pixels associated a plurality of objects of the particular type in the image. Additional image processing steps are developed to further refine the identified pixels for better fitting of the contour of the objects with their edges.

First claim

What is claimed is:

1. A computer-implemented method of classifying pixels associated with objects in an image, comprising:
receiving an image;
receiving training data, wherein the training data identifies a pixel associated with an object of a particular type in the image;
developing a plurality of filtered versions of the image;
processing the training data and the plurality of filtered versions of the image to develop a trained model for classifying pixels associated with objects of the particular type;
applying the trained model to the image to identify pixels associated with a plurality of objects of the particular type in the image; and
wherein the particular type is a biological cell and applying the trained model identifies pixels associated with boundaries between confluent biological cells in the image.

2. The computer-implemented method of claim 1 , further comprising receiving a further image and applying the trained model to the further image to identify pixels associated with a plurality of cells in the further image.

3. The computer-implemented method of claim 2 , wherein applying the trained model to the further image comprises developing a plurality of filtered versions of the further image.

4. The computer-implemented method of claim 3 , wherein:
developing each of the plurality of filtered versions of the image comprises applying a filter selected from a first set of filters;
developing each of the plurality of filtered versions of the further image comprises applying a filter selected from a second set of filters; and
wherein the first set of filters and the second set of filters are not identical.

5. The computer-implemented method of claim 1 , wherein applying the trained model includes using one of a Random Forest Decision Tree algorithm, a neural network, support vector machines, and a k-means clustering algorithm.

6. The computer-implemented method of claim 1 , wherein developing a filtered version of the image includes applying one of an edge detection filter, a peak detection filter, a subsampling filter, and a smoothing operator.

7. The computer-implemented method of claim 1 , further comprising determining how many cells are represented by the identified pixels associated with the plurality of cells in the image.

8. The computer-implemented method of claim 1 , further comprising growing an identified portion of an identified cell to select additional pixels associated with the cell.

9. A system for classifying pixels associated with objects in an image, wherein the system includes one or more processors, comprising:
a computer-based image acquisition module operating on the one or more processors and configured to receive an image from an image source;
a computer-based user interface module operating on the one or more processors and configured to receive training data, wherein the training data identifies a pixel associated with an object of a particular type in the image;
a computer-based filtering module operating on the one or more processors and configured to automatically develop a plurality of filtered versions of the image;
a computer-based training module operating on the one or more processors and configured to automatically process the training data and the plurality of filtered versions of the image to develop a trained model for classifying pixels associated with objects of the particular type;
a computer-based object identification module operating on the one or more processors and configured to automatically apply the trained model to the image to identify pixels associated with a plurality of objects of the particular type in the image; and
wherein the particular type is a biological cell and wherein the object identification module identifies pixels associated with boundaries between confluent biological cells in the image.

10. The system of claim 9 , wherein the computer-based image acquisition module is configured to receive a further image and the computer-based cell identification module is configured to apply the trained model to the further image to identify pixels associated with a plurality of cells in the further image.

11. The system of claim 10 , wherein the computer-based filtering module is configured to automatically develop filtered versions of the further image and the computer-based cell identification module is configured to automatically use the filtered versions of the further image to identify pixels associated with the plurality of cells in the image.

12. The system of claim 11 , wherein the computer-based filtering module is configured to automatically apply a first set of filters to develop the plurality of filtered versions of the image and to automatically use a second set of filters to develop the plurality of filtered versions of the further image, and wherein the first set of filters and the second set of filters are not identical.

13. The system of claim 9 , wherein the computer-based object identification module is configured to automatically use one of a Random Forest Decision Tree algorithm, a neural network, support vector machines, and a k-means clustering algorithm.

14. The system of claim 9 , wherein the computer-based filtering module is configured to automatically apply to the image one of an edge detection filter, a peak detection filter, a subsampling filter, and a smoothing operator.

15. The system of claim 9 , further comprising a computer-based cell measurement module operating on the one or more processors and configured to automatically determine how many cells are represented by the identified pixels associated with the plurality of cells in the image.

16. The system of claim 9 , wherein the computer-based cell measurement module is configured to automatically calculate statistics associated with the plurality of cells in the image.

Assignees

• Molecular Devices Llc
• Dh Technologies Dev Pte Ltd

Inventors

• Cohen Avrum Isaac
• Hong Dihui

Classifications

• G06V30/19167 Primary

  Active pattern learning · CPC title

• G06V20/698

  Matching; Classification · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/23

  Clustering techniques · CPC title

• G06F18/2178

  based on feedback of a supervisor · CPC title