Systems and methods for compound concentration sensing in fluids

Patent metadata

Field	Value
Publication number	US-12474258-B2
Application number	US-202217860412-A
Country	US
Kind code	B2
Filing date	Jul 8, 2022
Priority date	Jul 8, 2021
Publication date	Nov 18, 2025
Grant date	Nov 18, 2025

Abstract

A non-contact system for the sensing the concentration of a compound includes a hyperspectral imaging device configured to capture a hyperspectral image of a fluid, a flow cell configured to enable the capturing of a hyperspectral image of a fluid, a process, and a memory. The memory includes instructions stored thereon which, when executed by the processor, cause the system to generate a hyperspectral image of the fluid in the flow cell, generate several spectral signals based on the hyperspectral image, provide the spectral signal as an input to a machine learning network, and predict by the machine learning network the concentration of a compound in a fluid.

First claim

What is claimed is:

1 . A non-contact system for sensing a concentration of a compound, the system comprising:
a hyperspectral imaging device configured to capture a hyperspectral image of a fluid; a flow cell configured to enable the capturing of the hyperspectral image of the fluid; a processor; and a memory including instructions stored thereon which, when executed by the processor, cause the system to:
generate the hyperspectral image of the fluid in the flow cell;
generate a digital spectral signal based on the hyperspectral image, wherein the digital spectral signal includes one or more columns comprising pixel data;
provide the digital spectral signal as an input to a machine learning network;
predict by the machine learning network the concentration of the compound in the fluid;
compute an average of spectral intensity for each column of one or more columns in the hyperspectral image of the fluid, wherein the pixel data of each column includes a set of pixel intensities representing the spectral intensity at a specific wavelength across a spatial domain of a hyperspectral image;
generate baseline hyperspectral images of water and baseline hyperspectral images of glucose;
compute an average of spectral intensity for each column of pixel data in the baseline hyperspectral image of water;
compute an average of spectral intensity for each column of pixel data in the baseline hyperspectral image of glucose; and
determine a reference correction, by subtracting, for each column, the average spectral intensity from the baseline water hyperspectral image from the corresponding column average spectral intensity of the baseline glucose hyperspectral image.

2 . The system of claim 1 , further comprising a pump configured to pump the fluid into the flow cell.

3 . The system of claim 2 , further comprising a cell media filter configured to filter the fluid prior to the fluid being flowed into the flow cell.

4 . The system of claim 1 , wherein the machine learning network includes a convolutional neural network.

5 . The system of claim 1 , wherein the flow cell includes a transparent window configured for imaging the fluid.

6 . The system of claim 1 , wherein the instructions, when executed by the processor, further cause the system to preprocess the spectral signal to reduce noise, before providing the spectral signal to the machine learning network.

7 . The system of claim 1 , wherein the instructions, when executed by the processor, further cause the system to establish a relationship between the digital spectral signal and the concentration of the compound based on a ground truth of the concentration of the compound.

8 . The system of claim 7 , wherein the instructions, when executed by the processor, further cause the system to validate the relationship by leave-one-concentration-out (LOCO) cross validation.

9 . The system of claim 1 , wherein the hyperspectral imaging device is a short-wave infrared hyperspectral imaging device.

10 . A computer-implemented method for sensing a concentration of a compound, the method comprising:
capturing a hyperspectral image of a fluid within a flow cell by a hyperspectral imaging device; generating a spectral signal based on the hyperspectral image captured; providing the spectral signal as an input to a machine learning network; predicting by the machine learning network the concentration of the compound in the fluid; computing an average of spectral intensity for each column of pixel data in the hyperspectral image of the fluid; generating baseline hyperspectral images of water and baseline hyperspectral images of glucose; computing an average of spectral intensity for each column of pixel data in the baseline hyperspectral image of water; computing an average of spectral intensity for each column of pixel data in the baseline hyperspectral image of glucose; and determine a reference correction, by subtracting, for each column, the average spectral intensity from the baseline water hyperspectral image from the corresponding column average spectral intensity of the baseline glucose hyperspectral image.

11 . The computer-implemented method of claim 10 , further comprising pumping a filtered fluid into the flow cell by a pump.

12 . The computer-implemented method of claim 11 , further comprising filtering the fluid by a cell media filter prior to the flowing of the fluid through the flow cell.

13 . The computer-implemented method of claim 10 , wherein the machine learning network includes a convolutional neural network.

14 . The computer-implemented method of claim 10 , wherein the hyperspectral image is captured within a transparent window of the flow cell.

15 . The computer-implemented method of claim 10 , further comprising preprocessing the spectral signal to reduce noise, before providing the spectral signal to the machine learning network.

16 . The computer-implemented method of claim 10 , further comprising determining a relationship between the spectral signal and the concentration of the compound based on a ground truth of the concentration of the compound.

17 . A non-transitory computer-readable medium storing instructions which, when executed by a processor, cause the processor to perform a computer-implemented method for sensing a concentration of a compound, the method comprising:
generating a hyperspectral image of a fluid within a flow cell by a hyperspectral imaging device; generating a spectral signal based on the hyperspectral image captured; providing the digital spectral signal as an input to a machine learning network; predicting by the machine learning network a concentration of the compound in the fluid; computing an average of spectral intensity for each column of pixel data in the hyperspectral image of the fluid; generating baseline hyperspectral images of water and baseline hyperspectral images of glucose; computing an average of spectral intensity for each column of pixel data in the baseline hyperspectral image of water; computing an average of spectral intensity for each column of pixel data in the baseline hyperspectral image of glucose; and determine a reference correction, by subtracting, for each column, the average spectral intensity from the baseline water hyperspectral image from the corresponding column average spectral intensity of the baseline glucose hyperspectral image.

Assignees

• Univ Maryland

Inventors

• Tao Yang
• Wang Chiao-Yi
• Hevaganinge Anjana
• Wang Dongyi
• Ali Mohamed Amr

Classifications

• G01N2201/126

  Microprocessor processing · CPC title

• G01N21/359 Primary

  using near infrared light · CPC title

• G01J3/2823

  Imaging spectrometer · CPC title

• G01J3/108

  for measurement in the infrared range · CPC title

• G01J2003/2843

  Processing for eliminating interfering spectra · CPC title