Methods and apparatus for hashing and retrieval of training images used in HILN determinations of specimens in automated diagnostic analysis systems

Patent metadata

Field	Value
Publication number	US-12287320-B2
Application number	US-202017755467-A
Country	US
Kind code	B2
Filing date	Oct 22, 2020
Priority date	Oct 31, 2019
Publication date	Apr 29, 2025
Grant date	Apr 29, 2025

Abstract

A method of characterizing a specimen to be analyzed in an automated diagnostic analysis system provides an HILN classification (hemolysis, icterus, lipemia, normal) of the specimen along with a basis for that determination. The method includes assigning a hash code to each training image of a sample specimen used in the characterization training process. In response to an HILN determination for a test specimen, the method can retrieve via the hash code one or more of the closest matching training images upon which the HILN classification is based. The one or more of the closest matching training images can be displayed alongside of the one or more images of the test specimen. Quality check modules and systems configured to carry out the method are also described, as are other aspects.

First claim

What is claimed is:

1. A method of characterizing a specimen in an automated diagnostic analysis system, comprising:
receiving a plurality of training images to train an HILN network of a quality check module comprising a computer in the automated diagnostic analysis system, each of the training images depicting a sample specimen in a specimen container; and
assigning a hash code to each of the training images via a hashing network of the HILN network to facilitate retrieval of the one or more of the training images from a computer memory.

2. The method of characterizing a specimen of claim 1 , comprising:
receiving one or more images of a specimen in a specimen container to be analyzed in the automated diagnostic analysis system;
characterizing the specimen to be analyzed based on the one or more images via the HILN network using the plurality of training images to determine a classification index comprising hemolytic, icteric, lipemic, or normal classes.

3. The method of characterizing a specimen of claim 2 , comprising:
retrieving via the hash code of one or more of the plurality of training images upon which the classification index of the specimen is based.

4. The method of characterizing a specimen of claim 3 , wherein the retrieving via the hash code is conditioned on a determination that a classification index of the specimen is incorrect.

5. The method of characterizing a specimen of claim 4 , wherein the determination that the classification index of the characterized specimen is incorrect is based upon a confidence level.

6. The method of characterizing a specimen of claim 1 , comprising:
comparing one or more images of a specimen in a specimen container with an incorrect HILN determination to the one or more of the plurality of training images upon which the incorrect HILN determination is based.

7. The method of characterizing a specimen of claim 1 , wherein the plurality of training images comprise multi-viewpoint images captured by a plurality of image capture devices.

8. The method of characterizing a specimen of claim 1 , wherein the HILN network comprises a deep semantic segmentation network (DSSN).

9. The method of characterizing a specimen of claim 1 , wherein each of the plurality of training images comprises a classification that comprises one of a hemolytic class, an icteric class, a lipemic class, and a normal class.

10. The method of characterizing a specimen of claim 1 , wherein each of the plurality of training images includes a hemolytic sub-class, an icteric sub-class, or a lipemic sub-class.

11. The method of characterizing a specimen of claim 1 , wherein the hashing network comprises a front-end hashing network that assigns the hash code prior to segmentation.

12. The method of characterizing a specimen of claim 1 , wherein the hashing network comprises a back-end hashing network configured to receive a segmented region of each of the training images and assign the hash code to the segmented region of the specimen.

13. The method of characterizing a specimen of claim 1 , wherein assigned training image hash codes are stored in a database of a computer and later retrieved.

14. The method of characterizing a specimen of claim 1 , wherein the hashing network assigns the hash code to each group of training images representing a same sample specimen.

15. The method of characterizing a specimen of claim 1 , comprising retrieving via the hash code of one or more of the plurality of training images upon which classification of the specimen is based, and presenting the one or more of the plurality of training images to a user on a display screen.

16. The method of characterizing a specimen of claim 1 , comprising providing to the HILN network, one or more additional training images representing an incorrectly-characterized specimen.

17. The method of characterizing a specimen of claim 1 , wherein the method of characterizing is carried out by the quality check module.

18. The method of characterizing a specimen of claim 1 wherein hash codes for training images in a same HILN class have a first distance between them, while hash codes across HILN classes have larger distances between them.

19. A method of characterizing a specimen, comprising:
receiving a plurality of training images to train an HILN network, each of the training images depicting a sample specimen in a specimen container;
assigning a hash code to each of the training images via a hashing network;
receiving one or more images of a specimen in a specimen container to be analyzed by the HILN network;
characterizing the specimen to be analyzed based on the one or more images via the HILN network using the plurality of training images to determine a classification index of a hemolytic, icteric, lipemic, or normal class; and
retrieving, via the hash code, one or more of the plurality of training images upon which the classification index is based.

20. A quality check module of an automated diagnostic analysis system, comprising:
a plurality of image capture devices arranged around an imaging location and configured to capture multiple images from multiple viewpoints of a specimen container containing a specimen therein; and
a computer coupled to the plurality of image capture devices, the computer configured and operative via programming instructions to:
input a first plurality of images captured by the plurality of image capture devices to an HILN network executing on the computer, the first plurality of images representing a plurality of training images to train the HILN network, each of the training images depicting a sample specimen in a specimen container,
assign a hash code to each of the training images via a hashing network,
input one or more second images captured by the plurality of image capture devices to the HILN network executing on the computer, the one or more second images representing a test specimen in a specimen container to be analyzed in the automated diagnostic analysis system,
characterize the test specimen to be analyzed based on the one or more second images via the HILN network using the plurality of training images to determine a classification index comprising a hemolytic class, icteric class, lipemic class, or normal class, and
retrieve, via the hash code, one or more of the plurality of training images upon which the classification index of the test specimen is based.

Assignees

• Siemens Healthcare Diagnostics Inc

Inventors

• Narasimhamurthy Venkatesh
• Singh Vivek
• Chang Yao-Jen
• Pollack Benjamin S
• Kapoor Ankur
• Nalam Venkat Rayal Raj Prasad

Classifications

• G06T2207/30024

  Cell structures in vitro; Tissue sections in vitro · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/10024

  Color image · CPC title

• G06T7/0016

  involving temporal comparison · CPC title