Label-free digital brightfield analysis of nucleic acid amplification

1. An optical readout method for target nucleic acid detection comprising:

generating a plurality of droplets containing a loop-mediated isothermal amplification (LAMP) reaction mix, DNA primers specific to a target nucleic acid, and the target nucleic acid sample;

incubating the generated droplets;

imaging the incubated droplets using a brightfield imaging device to obtain one or more images;

subjecting the one or more images to image processing using image processing software executed on a computing device, wherein image processing comprises:

isolating individual droplets in the one or more images;

performing feature detection within the isolated droplets in the one or more images;

extracting keypoints and information related thereto from the detected features within the isolated droplets;

subjecting the extracted keypoints to a clustering operation to generate a plurality of words; and

inputting the word frequency into a trained machine learning droplet classifier, wherein the trained machine learning droplet classifier classifies each droplet as positive or negative.

2. The method of claim 1 , further comprising inputting total word count into the trained machine learning droplet classifier.

3. The method of claim 1 , further comprising inputting additional image features into the trained machine learning droplet classifier.

4. The method of claim 3 , wherein the additional image features comprise a compilation of strongly negative words.

5. The method of claim 3 , wherein the additional image features comprise a compilation of strongly positive words.

6. The method of claim 1 , wherein words are located in the center of the droplet are classified as likely precipitate words and words are located outside the center of the droplet are classified as likely non-precipitate words.

7. The method of claim 1 , wherein the additional image features comprise statistical information of the Laplacian of Gaussian Transformation of the one or more images.

8. The method of claim 1 , wherein the image processing software further outputs a concentration of the target nucleic acid based on the percentage or ratio of droplets that are classified as positive.

9. The method of claim 1 , wherein the keypoints are extracted using Speeded Up Robust Features (SURF) or Scale Invariant Feature Transform (SIFT).

10. The method of claim 1 , wherein the clustering comprises k-means clustering.

11. The method of claim 1 , wherein the trained machine learning droplet classifier classifies each droplet using one of Support Vector Machine (SVM), Random Forest, Adaptive Boosting, Joint Boost, or Logistic Regression.

12. An optical readout method for detecting a precipitate contained within a droplet comprising:

generating a plurality of droplets, at least some of the plurality of droplets comprising a precipitate generated during nucleic acid amplification contained therein;

imaging the droplets using a brightfield imaging device to obtain one or more images;

subjecting the one or more images to image processing using image processing software executed on a computing device, wherein image processing comprises:

isolating individual droplets in the one or more images;

performing feature detection within the isolated droplets in the one or more images;

extracting keypoints and information related thereto from the detected features within the isolated droplets;

subjecting the extracted keypoints to a clustering operation to generate a plurality of words; and

inputting the word frequency into a trained machine learning droplet classifier, wherein the trained machine learning droplet classifier classifies each droplet as positive for the precipitate or negative for the precipitate.

13. The method of claim 12 , further comprising inputting total word count into the trained machine learning droplet classifier.

14. The method of claim 12 , further comprising inputting additional image features into the trained machine learning droplet classifier.

15. The method of claim 14 , wherein the additional image features comprise a compilation of strongly negative words.

16. The method of claim 14 , wherein the additional image features comprise a compilation of strongly positive words.

17. The method of claim 12 , wherein words are located in the center of the droplet are classified as likely precipitate words and words located outside the center of the droplet are classified as likely non-precipitate words.

18. The method of claim 12 , wherein the additional image features comprise statistical information of the Laplacian of Gaussian Transformation of the one or more images.

19. A system for the optical readout of droplets containing a precipitate therein comprising:

a microfluidic device configured to generate a plurality of droplets, at some of the plurality of droplets comprising a precipitate generated during nucleic acid amplification contained therein;

a brightfield imaging device configured to obtain an image of a field of view (FOV) containing the plurality of droplets;

a computing device configured to execute image processing software, wherein image processing software is configured to:

isolate individual droplets in the image;

perform feature detection within the individual droplets in the image;

extract keypoints and information related thereto from the detected features within the individual droplet;

clustering the keypoints to generate a plurality of words; and

input the word frequency into a trained machine learning droplet classifier executed by the image processing software, wherein the trained machine learning droplet classifier classifies each droplet as positive for the precipitate or negative for the precipitate.