Methods and apparatus for classifying an artifact in a specimen

Patent metadata

Field	Value
Publication number	US-10746665-B2
Application number	US-201716072386-A
Country	US
Kind code	B2
Filing date	Jan 24, 2017
Priority date	Jan 28, 2016
Publication date	Aug 18, 2020
Grant date	Aug 18, 2020

Abstract

A model-based method of inspecting a specimen for presence of one or more artifacts (e.g., a clot, bubble, and/or foam). The method includes capturing multiple images of the specimen at multiple different exposures and at multiple spectra having different nominal wavelengths, selection of optimally-exposed pixels from the captured images to generate optimally-exposed image data for each spectra, computing statistics of the optimally-exposed pixels to generate statistical data, identifying a serum or plasma portion of the specimen, and classifying, based on the statistical data, whether an artifact is present or absent within the serum or plasma portion. Testing apparatus and quality check modules adapted to carry out the method are described, as are other aspects.

First claim

What is claimed is:

1. A method of determining an artifact in a specimen contained within a specimen container, comprising:
providing a specimen that is separated and contained in a specimen container;
capturing images of the specimen at multiple different exposures and at multiple different wavelengths;
selecting optimally-exposed pixels from the captured images at the different exposures at each wavelength to generate optimally-exposed image data for each wavelength;
computing statistics on the optimally-exposed pixels at the different wavelengths to generate statistical data;
identifying a serum or plasma portion of the specimen based on the statistical data; and
classifying, based on the statistical data, whether an artifact is:
present within one or more regions of the serum or plasma portion, or
absent within the serum or plasma portion.

2. The method of claim 1 , wherein the artifact is selected from a group including at least one of a clot, a bubble, or foam.

3. The method of claim 1 , wherein the specimen is centrifuged and includes a separated blood portion and a serum or plasma portion.

4. The method of claim 1 , wherein the capturing images of the specimen involves capturing multiple images taken from a number of viewpoints.

5. The method of claim 4 , wherein the number of viewpoints comprises 3 or more.

6. The method of claim 1 , wherein the providing the specimen comprises securing the specimen container containing the specimen in a holder.

7. The method of claim 1 , wherein the multiple different wavelengths comprise at least two wavelengths between about 400 nm and about 700 nm.

8. The method of claim 1 , wherein the multiple wavelengths comprise at least two wavelengths selected from a group of about 455 nm, about 537 nm, and about 634 nm.

9. The method of claim 1 , wherein the multiple exposures times comprise between about 0.1 ms and about 256 ms.

10. The method of claim 1 , wherein the selecting optimally-exposed pixels comprises selecting pixels from the images that include intensities of between about 180-254 based upon an intensity range of 0-255.

11. The method of claim 1 , wherein the identifying a serum or plasma portion is based upon classifying of pixels in optimally-exposed image data based upon a multi-class classifier generated from multiple training sets.

12. The method of claim 11 , wherein the multi-class classifier further comprises a support vector machine or a random decision tree.

13. The method of claim 1 , wherein classifying, based on the statistical data whether an artifact is present within one or more regions of the serum or plasma portion, or is absence within the serum or plasma portion, is based upon one or more classifiers generated from multiple artifact training sets.

14. The method of claim 13 , wherein the one or more classifiers comprises a separate binary classifier for each of clot, bubble, and foam.

15. The method of claim 1 , wherein the computing statistics of the optimally-exposed pixels from the optimally-exposed image data for each wavelength comprises calculating a mean value, a standard deviation, and/or covariance from a collection of corresponding pixels from each wavelength.

16. The method of claim 1 , comprising determining an identity of the specimen based upon deciphering a bar code in the capturing images.

17. The method of claim 1 , comprising grouping the optimally-exposed image data of the specimen to form a 3D model.

18. The method of claim 1 , wherein the identifying a serum or plasma portion of the separated specimen comprises ignoring portions of a holder within the images.

19. A quality check module adapted to determine presence of an artifact in a specimen contained within a specimen container, comprising:
a plurality of cameras arranged around the specimen container and configured to capture multiple images of the specimen container from multiple viewpoints, each adapted to generate a plurality of images taken at multiple different exposure times and at multiple different wavelengths; and
a computer coupled to the plurality of cameras and adapted to process image data from the images, the computer configured and capable of being operated to:
select optimally-exposed pixels from the images at the different exposure times to generate optimally-exposed image data for each wavelength,
compute statistics of the optimally-exposed pixels at each of the wavelengths to generate statistical data,
identify a serum or plasma portion of the specimen, and

classify, based on the statistical data, whether an artifact is:
present within one or more regions of the serum or plasma portion, or
absent within the serum or plasma portion.

20. A testing apparatus adapted to determine a presence of an artifact in a specimen contained within a specimen container, comprising:
a track;
a carrier on the track that is configured to contain the specimen container;
a plurality of cameras arranged around the track and configured to capture multiple images of the specimen container from multiple viewpoints, each camera is configured to generate a plurality of images at multiple different exposures and multiple different wavelengths; and
a computer coupled to the cameras and configured to process image data from the multiple images, the computer configured and capable of being operated to:
select optimally-exposed pixels from the multiple images at the different exposures to generate optimally-exposed image data for each wavelength,
compute statistics on the optimally-exposed pixels at the different wavelengths to generate statistical data,
identify a serum or plasma portion of the specimen, and
classify, based on the statistical data, whether an artifact is:
present within one or more regions of the serum or plasma portion, or
absent within the serum or plasma portion.

Assignees

• Siemens Healthcare Diagnostics Inc

Inventors

• Kluckner Stefan
• Chang Yao-Jen
• Chen Terrence
• Pollack Benjamin S

Classifications

• G06T7/0012 Primary

  Biomedical image inspection · CPC title

• G01N21/9036

  using arrays of emitters or receivers · CPC title

• G01N2035/1018

  Detecting inhomogeneities, e.g. foam, bubbles, clots · CPC title

• G01N21/9027

  in containers after filling · CPC title

• G06T7/143

  involving probabilistic approaches, e.g. Markov random field [MRF] modelling · CPC title