Microorganism information providing device and method

Patent metadata

Field	Value
Publication number	US-12228509-B2
Application number	US-201917273631-A
Country	US
Kind code	B2
Filing date	Sep 6, 2019
Priority date	Sep 7, 2018
Publication date	Feb 18, 2025
Grant date	Feb 18, 2025

Abstract

According to an embodiment of the present disclosure, provided is an apparatus for providing microorganism information, including: a receiving unit configured to receive a plurality of images obtained by photographing in time series an outgoing wave emitted from a sample; a detecting unit configured to extract a feature of a change over time from the plurality of images obtained by photographing in time series; a learning unit configured to machine-learn classification criteria based on the extracted feature; and a determining unit configured to classify the type or concentration of a microorganism included in the sample based on the classification criteria, wherein each of the plurality of images includes speckle information generated by multiple scattering by the microorganism due to waves incident on the sample.

First claim

What is claimed is:

1. An apparatus for providing microorganism information, the apparatus comprising:
a receiving unit configured to receive a plurality of images obtained by photographing in time series an outgoing wave emitted from a sample;
a detecting unit configured to extract a feature of a change over time from the plurality of images obtained in time series;
a learning unit configured to machine-learn classification criteria based on an extracted feature; and
a determining unit configured to classify the type or concentration of a microorganism contained in the sample based on the classification criteria, wherein
each of the plurality of images includes speckle information generated by multiple scattering caused by the microorganism due to a wave entering into the sample,
wherein the learning unit learns the classification criteria by using a convolutional neural network (CNN), and
wherein the learning unit performs a convolution arithmetic by using a convolution kernel having a size smaller than a size of one speckle.

2. The apparatus of claim 1 , wherein
the learning unit performs a convolution arithmetic by using a convolution kernel having a size of n×n smaller than m when a size of one speckle corresponds to m pixels.

3. The apparatus of claim 2 , wherein
a stride of the convolution arithmetic corresponds to the value n.

4. The apparatus of claim 2 , wherein
the convolution kernel comprises kernel feature maps corresponding to or greater than the number of the plurality of images.

5. The apparatus of claim 4 , wherein
the learning unit performs a convolution arithmetic by using a kernel set including a plurality of convolution kernels corresponding to the number of output channels.

6. The apparatus of claim 5 , wherein
the number of output channels by the convolution arithmetic corresponds to the number of the plurality of images.

7. The apparatus of claim 1 , wherein
the classification criteria is learned by using one of a change in a shape of a speckle pattern in the feature, a temporal correlation coefficient calculated based on an intensity of light of the speckle pattern, and a change in a standard deviation value of the intensity of light of the speckle pattern.

8. The apparatus of claim 7 , wherein
the standard deviation value and the concentration of the microorganism has a linear relationship.

9. An apparatus for providing impurity information, the apparatus comprising:
a receiving unit configured to receive a plurality of images obtained by photographing in time series an outgoing wave emitted from a sample;
a detecting unit configured to extract a feature of a change over time from the plurality of images obtained in time series;
a learning unit configured to machine-learn classification criteria based on an extracted feature; and
a determining unit configured to classify the type or concentration of an impurity contained in the sample based on the classification criteria, wherein
each of the plurality of images includes speckle information generated by multiple scattering caused by the impurity due to a wave entering into the sample, and
the learning unit is configured to distinguish one speckle from surrounding speckles in the plurality of images and to learn the classification criteria using time information of the one speckle.

10. A method of providing microorganism information, the method comprising:
receiving a plurality of training images which have been obtained by photographing in time series an outgoing wave emitted by irradiating a wave to a sample containing a microorganism of which type or concentration is previously known;
machine-learning classification criteria based on a feature of a change over time from the plurality of training images which have been obtained in time series;
receiving a plurality of images obtained by photographing in time series an outgoing wave which is emitted by irradiating a wave onto a new sample; and
identifying a type or a concentration of a microorganism included in the new sample based on the plurality of images and the classification criteria, wherein
each of the plurality of training images or each of the plurality of images comprises speckle information that is generated by multiple scattering by the microorganism due to waves incident on the sample or the new sample,
wherein
the machine-learning of the classification criteria comprises
learning the classification criteria by using a convolutional neural network (CNN) and
performing a convolution arithmetic by using a convolution kernel having a size smaller than a size of one speckle.

11. The method of claim 10 , wherein
the machine-learning of the classification criteria comprises
performing a convolution arithmetic by using a convolution kernel having a size of n×n smaller than m when a size of one speckle corresponds to m pixels.

12. The method of claim 11 , wherein
a stride of the convolution arithmetic corresponds to the value n.

13. The method of claim 10 , wherein the convolution kernel comprises kernel feature maps corresponding to or greater than the number of the plurality of images.

14. The method of claim 13 , wherein
the machine-learning of the classification criteria comprises
performing a convolution arithmetic by using a kernel set including a plurality of convolution kernels corresponding to the number of output channels.

15. The method of claim 14 , wherein
the number of output channels by the convolution arithmetic corresponds to the number of the plurality of images.

16. The method of claim 10 , wherein
the learning unit learns the classification criteria based on temporal correlation of the plurality of images.

Assignees

• The Wave Talk Inc
• Korea Advanced Inst Sci & Tech

Inventors

• Park Yongkeun
• Lee KyeoReh
• Shin Seungwoo
• Kim Geon
• Kim Young Dug

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06V20/698

  Matching; Classification · CPC title

• G06V20/695

  Preprocessing, e.g. image segmentation · CPC title

• G06V10/82

  using neural networks · CPC title