AI-driven enhancement of motion blurred sequencing images
Artificial intelligence driven enhancement of motion blurred sequencing images enables enhanced sequencing that determines a sequence of bases in genetic material with any one or more of: improved performance, improved accuracy, and/or reduced cost. A training set of images taken after unreduced and reduced movement settling times during sequencing is used to train a neural network to enable the neural network to recover enhanced images, as if taken after the unreduced movement settling time, from unenhanced images taken after the reduced movement settling time.
1 . A method of reducing settling time effects in images used for collected images during sequencing, the method comprising:
accessing a training set of paired images taken after an unreduced settling time and after a reduced settling time during a sequencing operation;
wherein a settling time reduction ratio between the unreduced settling time, before reduction, and the reduced settling time, after reduction, is at least 2 to 1;
training a convolutional neural network comprising a generative adversarial network that has a generator stage and a discriminator stage each updating respective pluralities of parameters during the training, the plurality of parameters of the generator stage enabling recovering enhanced images, from unenhanced images taken after the reduced settling time;
whereby trained filters of the convolutional neural network enable adding information to images taken after the reduced settling time to enable production of the enhanced images; and
saving the trained filters for use in processing collected images from sequencing taken after the reduced settling time.
2 . The method of claim 1 , wherein the generator stage provides candidate enhanced images to the discriminator stage.
3 . The method of claim 1 , wherein the convolutional neural network is a training convolutional neural network, and the generator stage is a training generator stage, and further comprising accessing production images taken after the reduced settling time and using information of the trained filters in a production convolutional neural network that has a production generator stage to enhance the production images as if taken after the unreduced settling time.
4 . The method of claim 1 , further comprising producing the images of the training set taken after the unreduced settling time by imaging after cessation of vibrations between an imager and samples and further comprising producing the images of the training set taken after the reduced settling time by imaging before the cessation of vibrations.
5 . The method of claim 1 , further comprising base calling from the enhanced images.
6 . The method of claim 1 , wherein the convolutional neural network comprises any combination of any one or more of
one or more 1D convolutional layers,
one or more 2D convolutional layers,
one or more 3D convolutional layers,
one or more 4D convolutional layers,
one or more 5D convolutional layers,
one or more multi-dimensional convolutional layers,
one or more single channel convolutional layers,
one or more multi-channel convolutional layers,
one or more 1×1 convolutional layers,
one or more atrous convolutional layers,
one or more transpose convolutional layers,
one or more depthwise separable convolutional layers,
one or more pointwise convolutional layers,
one or more 1×1 convolutional layers,
one or more group convolutional layers,
one or more flattened convolutional layers,
one or more spatial convolutional layers,
one or more spatially separable convolutional layers,
one or more cross-channel convolutional layers,
one or more shuffled grouped convolutional layers,
one or more pointwise grouped convolutional layers,
one or more upsampling layers,
one or more downsampling layers,
one or more averaging layers, and
one or more padding layers.
7 . The method of claim 1 , wherein the training comprises determining one or more loss terms comprising any combination of any one or more of a logistic regression/log loss, a multi-class cross-entropy/softmax loss, a binary cross-entropy loss, a mean squared error loss, a mean absolute error loss, a mean absolute percentage error loss, a mean squared logarithmic error loss, an L 1 loss, an L 2 loss, a smooth L 1 loss, a Huber loss, a patch-based loss, a pixel-based loss, a pixel-wise loss, a single-image loss, adversarial loss, and a fiducial-based loss.
8 . The method of claim 1 , wherein each of the images taken after the reduced settling time is produced by capturing multiple images of a same tile with a multiple capture imager and then processing the multiple images with an AI model to produce the respective image taken after the reduced settling time.
9 . The method of claim 1 , wherein the convolutional neural network is a training convolutional neural network comprised in a training sequencing instrument and further comprising training a production convolutional neural network comprised in a production sequencing instrument, the training the production convolutional neural network starting with information of the trained filters and updating parameters of the production convolutional neural network based on processing fiducial elements of tuning images obtained via the production sequencing instrument and wherein the tuning images are taken after the reduced settling time.
10 . The method of claim 1 , wherein the sequencing is sequencing by synthesis and comprises a plurality of cycles, each cycle corresponding to a single base call for each of a plurality of oligos, each cycle occurring one after another sequentially, and the training is performed with respect to a plurality of contiguous non-overlapping ranges of the cycles, resulting in a plurality of trained filters each corresponding to a respective one of the non-overlapping cycle ranges.
11 . The method of claim 1 , further comprising determining image quality of the enhanced images, and responsive to the quality being below a threshold, recapturing one or more of the images taken after the reduced settling time using the unreduced settling time.
12 . The method of claim 1 , further comprising pretraining the convolutional neural network using pretraining images taken after a settling time that is greater than the reduced settling time and less than the unreduced settling time.
13 . The method of claim 1 , wherein the training set comprises information regarding a respective x-y stage position of each image.
14 . The method of claim 1 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at every cycle of the plurality of imaging cycles.
15 . The method of claim 1 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at less than every cycle of the plurality of imaging cycles.
16 . A method of reducing settling time used for collecting images during a sequencing operation, the method comprising:
accessing a training set of paired images taken after an unreduced settling time and after a reduced settling time during sequencing, the reduced settling time being before the unreduced settling time;
wherein a settling time reduction ratio between the unreduced settling time, before reduction, and the reduced settling time, after reduction, is at least 2 to 1;
training a convolutional neural network that has an encoder stage and a decoder stage each updating respective pluralities of parameters during the training, the respective pluralities of parameters collectively enabling substantially recovering enhanced images, as if taken after the unreduced settling time, from unenhanced images taken after the reduced settling time, after reduction;
whereby trained filters of the convolutional neural network enable adding information to images taken after the reduced settling time to enable production of the enhanced images; and
saving the trained filters for use processing collected images from sequencing taken after the reduced settling time.
17 . The method of claim 16 , wherein the encoder stage provides an intermediate representation to the decoder stage.
18 . The method of claim 16 , wherein the convolutional neural network further comprises one or more skip connections between the encoder and decoder stages.
19 . The method of claim 16 , wherein the convolutional neural network is a training convolutional neural network and further comprising accessing production images taken after the reduced settling time and using information of the trained filters in a production convolutional neural network to enhance the production images as if taken after the unreduced settling time.
20 . The method of claim 16 , further comprising base calling from the enhanced images.
21 . The method of claim 16 , wherein each of the images taken after the reduced settling time is produced by capturing multiple images of a same tile with a multiple capture imager and then processing the multiple images with an AI model to produce the respective image taken after the reduced settling time.
22 . The method of claim 16 , wherein the convolutional neural network is a training convolutional neural network comprised in a training sequencing instrument and further comprising training a production convolutional neural network comprised in a production sequencing instrument, the training the production convolutional neural network starting with information of the trained filters and updating parameters of the production convolutional neural network based on processing fiducial elements of tuning images obtained via the production sequencing instrument and wherein the tuning images are taken after the reduced settling time.
23 . The method of claim 16 , wherein the sequencing is sequencing by synthesis and comprises a plurality of cycles, each cycle corresponding to a single base call for each of a plurality of oligos, each cycle occurring one after another sequentially, and the training is performed with respect to a plurality of contiguous non-overlapping ranges of the cycles, resulting in a plurality of trained filters each corresponding to a respective one of the non-overlapping cycle ranges.
24 . The method of claim 16 , further comprising determining image quality of the enhanced images, and responsive to the quality being below a threshold, recapturing one or more of the images taken after the reduced settling time using the unreduced settling time.
25 . The method of claim 16 , further comprising pretraining the convolutional neural network using pretraining images taken after a settling time that is greater than the reduced settling time and less than the unreduced settling time.
26 . The method of claim 16 , wherein the training set comprises information regarding a respective x-y stage position of each image.
27 . The method of claim 16 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at every cycle of the plurality of imaging cycles.
28 . The method of claim 16 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at less than every cycle of the plurality of imaging cycles.
29 . A method of reducing settling time used for collecting images during a sequencing operation, the method comprising:
accessing a training set of images taken after an unreduced settling time and after a reduced settling time during sequencing;
wherein a settling time reduction ratio between the unreduced settling time, before reduction, and the reduced settling time, after reduction, is at least 2 to 1;
training a convolutional neural network comprising a cycle-consistent generative adversarial network that has first and second generator stages and first and second discriminator stages each of the generator stages and each of the discriminator stages updating respective pluralities of parameters during the training, the plurality of parameters of the first generator stage enabling substantially recovering enhanced images, as if taken after the unreduced settling time, from unenhanced images taken after the reduced settling time, after reduction;
whereby trained filters of the convolutional neural network enable adding information to images taken after the reduced settling time to enable production of the enhanced images; and
saving the trained filters for use processing collected images from sequencing taken after the reduced settling time.
30 . The method of claim 29 , further comprising base calling from the enhanced images.
31 . The method of claim 29 , wherein the training set comprises information regarding a respective x-y stage position of each image.
32 . The method of claim 29 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at every cycle of the plurality of imaging cycles.
33 . The method of claim 29 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at less than every cycle of the plurality of imaging cycles.
34 . A non-transitory computer readable storage medium impressed with computer program instructions, which, when executed on a processor, implement actions comprising:
accessing a training set of paired images taken after an unreduced settling time and after a reduced settling time during sequencing;
wherein a settling time reduction ratio between the unreduced settling time, before reduction, and the reduced settling time, after reduction, is at least 2 to 1;
training a convolutional neural network comprising a generative adversarial network that has a generator stage and a discriminator stage each updating respective pluralities of parameters during the training, the plurality of parameters of the generator stage enabling recovering enhanced images, from unenhanced images taken after the reduced settling time;
whereby trained filters of the convolutional neural network enable adding information to images taken after the reduced settling time to enable production of the enhanced images; and
saving the trained filters for use in processing collected images from sequencing taken after the reduced settling time.
35 . The non-transitory computer readable storage medium of claim 34 , wherein the actions further comprise base calling from the enhanced images.
36 . The non-transitory computer readable storage medium of claim 34 , wherein the training set comprises information regarding a respective x-y stage position of each image.
37 . The non-transitory computer readable storage medium of claim 34 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at every cycle of the plurality of imaging cycles.
38 . The non-transitory computer readable storage medium of claim 34 , wherein the sequencing operation has a plurality of imaging cycles, and wherein the paired images are taken at less than every cycle of the plurality of imaging cycles.
39 . A method of reducing settling time effects in images used for collected images during sequencing, the method comprising:
accessing a training set of paired images taken after an unreduced settling time and after a reduced settling time during a sequencing operation;
wherein a settling time reduction ratio between the unreduced settling time, before reduction, and the reduced settling time, after reduction, is at least 2 to 1;
training a convolutional neural network comprising two or more stages, each stage updating respective pluralities of parameters during the training, the plurality of parameters enabling substantially recovering enhanced images, from unenhanced images taken after the reduced settling time;
whereby trained filters of the convolutional neural network enable adding information to images taken after the reduced settling time to enable production of the enhanced images; and
saving the trained filters for use in processing collected images from sequencing taken after the reduced settling time.