Granted Patent B2

US 12,423,815 · App. 17/858,268 · Granted Sep 23, 2025

Inventor: Mohsen Rezaei (Brown Deer, WI)

Assignee: ILLUMINA, INC.

G06T7/0014G01N33/582G06T5/73G16B30/00G06T2207/20224

Abstract

Claims (93)

1. A method comprising:

obtaining a plurality of analysis images of light emitted during sequencing of a biological sample;

obtaining noise dependencies by performing acts comprising, for each location in a point spread function which comprises a plurality of locations, calculating a noise correlation for that location, wherein the noise correlation is a correlation between noise in a first plurality of measurements and noise in a second plurality of measurements, wherein:

each measurement from the first plurality of measurements is captured by a sensor whose position relative to a corresponding sensor which captured a measurement from the second plurality of measurements is the same as that location's position in the point spread function relative to a center of the point spread function; and

each measurement from the first plurality of measurements and the second plurality of measurements measures light emitted during sequencing of the biological sample from the plurality of analysis images;

populating the point spread function based on the noise dependencies;

generating a sharpening kernel based on the point spread function;

obtaining a plurality of sharpened images by applying the sharpening kernel to the plurality of analysis images;

one or more times, repeating:

obtaining noise dependencies;

populating the point spread function;

generating the sharpening kernel; and

obtaining the plurality of sharpened images;

wherein, on each repetition, the plurality of analysis images for that repetition of obtaining noise dependencies is the plurality of sharpened images from a most recent preceding application of the sharpening kernel;

identifying a sharpening kernel generated on a repetition as an optimal sharpening kernel; and

applying the optimal sharpening kernel to compensate for crosstalk in images subsequently captured while sequencing the biological sample.

2. The method of claim 1 , wherein repeating populating the point spread function comprises:

generating a dependency matrix having dimensions equal to those of the point spread function by, for each location in the point spread function, populating a corresponding location in the dependency matrix with a most recently obtained dependency for that location;

multiplying the dependency matrix by a scalar constant; and

adding a result of multiplying the dependency matrix by the scalar constant to a most recently populated preceding point spread function.

3. The method of claim 2 , wherein the scalar constant has a value greater than or equal to 0.08, and less than or equal to 0.12.

4. The method of claim 1 , wherein identifying the optimal sharpening kernel comprises:

determining, for each repetition, a signal to noise ratio obtained by applying the sharpening kernel generated on that repetition to the plurality of analysis images; and

identifying the sharpening kernel from which a highest signal to noise ratio is obtained as the optimal sharpening kernel.

5. The method of claim 4 , wherein determining, for each repetition, the signal to noise ratio comprises calculating a sharpness of the plurality of sharpened images obtained on that repetition.

6. The method of claim 1 , wherein obtaining noise dependencies comprises obtaining a noise map based on subtracting a first image from the plurality of analysis images from a second image from the plurality of analysis images.

7. The method of claim 6 , wherein:

obtaining noise dependencies comprises dividing the noise map into a plurality of units, wherein each unit is a matrix of values having dimensions at least as great as those of the point spread function;

for each location in the point spread function, obtaining the noise dependency for that location comprises calculating a correlation between a first set of values and a second set of values, wherein:

the first set of values comprises, for each unit from the plurality of units, a value at a first location in that unit; and

the second set of values comprises, for each unit from the plurality of units, a value at a second location in that unit, wherein the first location for the value from that unit in the first set of values has a position relative to the second location that is the same as that location's position in the point spread function relative to the center of the point spread function.

8. The method of claim 6 , wherein:

the first image from the plurality of analysis images is an image captured from a first sequencing cycle; and

the second image from the plurality of analysis images is an image from a second sequencing cycle.

9. The method of claim 1 , wherein:

each image from the plurality of analysis images comprises an image from a different sequencing cycle;

the plurality of analysis images comprises more than two images;

obtaining noise dependencies comprises:

obtaining a plurality of intermediate correlations, wherein each of the intermediate correlations corresponds to two analysis images from the plurality of analysis images, and wherein each of the plurality of intermediate correlations is obtained based on:

obtaining an intermediate map for that intermediate correlation by subtracting one of the analysis images corresponding to that intermediate correlation from the other analysis image corresponding to that intermediate correlation;

dividing the intermediate map for that intermediate correlation into a plurality of units, wherein each unit is a matrix of values having dimensions at least as great as those of the point spread function;

for each location in the point spread function, calculating a correlation between a first set of values and a second set of values, wherein:

the first set of values comprises, for each unit from the plurality of units, a value at a first location in that unit; and

the second set of values comprises, for each unit from the plurality of units, a value at a second location in that unit, wherein the first location for the value from that unit in the first set of values has a position relative to the second location that is the same as that location's position in the point spread function relative to the center of the point spread function;

determining the set of noise dependencies based on the plurality of intermediate correlations.

10. The method of claim 1 , wherein obtaining noise dependencies, populating the point spread function, generating the sharpening kernel, and obtaining the plurality of sharpened images are repeated between two and eight times.

11. A system comprising:

a sensor array;

a processor to:

obtain a plurality of analysis images of light emitted during sequencing of a biological sample;

obtain noise dependencies by performing acts comprising, for each location in a point spread function which comprises a plurality of locations, calculating a noise correlation for that location, wherein the noise correlation is a correlation between noise in a first plurality of measurements and noise in a second plurality of measurements, wherein:

each measurement from the first plurality of measurements is captured by a sensor from the sensor array whose position relative to a corresponding sensor in the sensor array which captured a measurement from the second plurality of measurements is the same as that location's position in the point spread function relative to a center of the point spread function; and

each measurement from the first plurality of measurements and the second plurality of measurements measures light emitted during sequencing of the biological sample from the plurality of analysis images;

populate the point spread function based on the noise dependencies;

generate a sharpening kernel based on the point spread function;

obtain a plurality of sharpened images by applying the sharpening kernel to the plurality of analysis images;

one or more times, repeat:

obtaining noise dependencies;

populating the point spread function;

generating the sharpening kernel; and

obtaining the plurality of sharpened images;

wherein, on each repetition, the plurality of analysis images for that repetition of obtaining noise dependencies is the plurality of sharpened images from a most recent preceding application of the sharpening kernel;

identify a sharpening kernel generated on a repetition as an optimal sharpening kernel; and

apply the optimal sharpening kernel to compensate for crosstalk in images subsequently captured while sequencing the biological sample.

12. The system of claim 11 , wherein repeating populating the point spread function comprises:

generating a dependency matrix having dimensions equal to those of the point spread function by, for each location in the point spread function, populating a corresponding location in the dependency matrix with a most recently obtained dependency for that location;

multiplying the dependency matrix by a scalar constant; and

adding a result of multiplying the dependency matrix by the scalar constant to a most recently populated preceding point spread function.

13. The system of claim 12 , wherein the scalar constant has a value greater than or equal to 0.08, and less than or equal to 0.12.

14. The system of claim 11 , wherein identifying the optimal sharpening kernel comprises:

determining, for each repetition, a signal to noise ratio obtained by applying the sharpening kernel generated on that repetition to the plurality of analysis images; and

identifying the sharpening kernel from which a highest signal to noise ratio is obtained as the optimal sharpening kernel.

15. The system of claim 14 , wherein determining, for each repetition, the signal to noise ratio comprises calculating a sharpness of the plurality of sharpened images obtained on that repetition.

16. The system of claim 11 , wherein obtaining noise dependencies comprises obtaining a noise map based on subtracting a first image from the plurality of analysis images from a second image from the plurality of analysis images.

17. The system of claim 16 , wherein:

obtaining noise dependencies comprises dividing the noise map into a plurality of units, wherein each unit is a matrix of values having dimensions at least as great as those of the point spread function;

for each location in the point spread function, obtaining the noise dependency for that location comprises calculating a correlation between a first set of values and a second set of values, wherein:

the first set of values comprises, for each unit from the plurality of units, a value at a first location in that unit; and

the second set of values comprises, for each unit from the plurality of units, a value at a second location in that unit, wherein the first location for the value from that unit in the first set of values has a position relative to the second location that is the same as that location's position in the point spread function relative to the center of the point spread function.

18. The system of claim 16 , wherein:

the first image from the plurality of analysis images is an image from a first sequencing cycle; and

the second image from the plurality of analysis images is an image from a second sequencing cycle.

19. The system of claim 11 , wherein:

each image from the plurality of analysis images comprises an image from a different sequencing cycle;

the plurality of analysis images comprises more than two images;

obtaining noise dependencies comprises:

obtaining a plurality of intermediate correlations, wherein each of the intermediate correlations corresponds to two analysis images from the plurality of analysis images, and wherein each of the plurality of intermediate correlations is obtained based on:

obtaining an intermediate map for that intermediate correlation by subtracting one of the analysis images corresponding to that intermediate correlation from the other analysis image corresponding to that intermediate correlation;

dividing the intermediate map for that intermediate correlation into a plurality of units, wherein each unit is a matrix of values having dimensions at least as great as those of the point spread function;

for each location in the point spread function, calculating a correlation between a first set of values and a second set of values, wherein:

the first set of values comprises, for each unit from the plurality of units, a value at a first location in that unit; and

the second set of values comprises, for each unit from the plurality of units, a value at a second location in that unit, wherein the first location for the value from that unit in the first set of values has a position relative to the second location that is the same as that location's position in the point spread function relative to the center of the point spread function;

determining the set of noise dependencies based on the plurality of intermediate correlations.

Assignments (1)

Continuity (2)

References Cited (62)