Verily Life Sciences LLC

Cheng-Hsun Wu

Supriyo Sinha

Ali Behrooz

James Michael Higbie

Charles Santori

Seung Ah Lee

John D Perreault

Chinmay Belthangady

Brian M Rabkin

Andrew Homyk

Victor Marcel Acosta

Jerrod Joseph Schwartz

Ian Peikon

Paul M Lebel

Huang-Wei Chang

Lan Peikon

Systems and methods for using a prediction model jointly with a normalization model to provide prediction results are provided. One example method includes receiving an input image of a tissue sample of a patient and generating a normalized image by applying a normalization model on the input image. The normalization model is configured to generate normalized data using input data for a prediction model, and the prediction model is configured to generate prediction results using normalized data generated by the normalization model. The normalization model and the prediction model are jointly trained. The method further includes generating a prediction of disease severity for the patient by applying the prediction model on the normalized image.

Auto-normalization for machine learning

One disclosed example method for simulated medical images using GANs includes receiving, by a generative adversarial network ('GAN'), a plurality of training images, the training images associated with an affliction and depicting different stages of progression for the affliction; generating, using the GAN, a latent space based on the training images, the latent space comprising a first set of data points indicating parameters associated with the training images; generating, using the GAN, a second set of data points in the latent space based on simulated images generated from the latent space, the simulated images based at least in part on the first set of data points; after generating the second set of data points: receiving a request for a first simulated image associated with the affliction; and generating and outputting, by the GAN, the first simulated image based on the latent space.

Systems and methods for simulating medical images using generative adversarial networks

Systems and methods for predicting images with enhanced spatial resolution using a neural network are provided herein. According to an aspect of the invention, a method includes accessing an input image of a biological sample, wherein the input image includes a first spatial resolution and a plurality of spectral images, and wherein each spectral image of the plurality of spectral images includes data from a different wavelength band at a different spectral channel; applying a trained artificial neural network to the input image; generating an output image at a second spatial resolution, wherein the second spatial resolution is higher than the first spatial resolution, and wherein the output image includes a fewer number of spectral channels than the plurality of spectral images included in the input image; and outputting the output image.

Transforming multispectral images to enhanced resolution images enabled by machine learning

Systems and methods of improving alignment in dense prediction neural networks are disclosed. A method includes identifying, at a computing system, an input data set and a label data set with one or more first parts of the input data set corresponding to a label. The computing system processes the input data set using a neural network to generate a predicted label data set that identifies one or more second parts of the input data set predicted to correspond to the label. The computing system determines an alignment result using the predicted label data set and the label data set and a transformation of the one or more first parts, including a shift, rotation, scaling, and/or deformation, based on the alignment result. The computing system computes a loss score using the transformation, label data and the predicted label data set and updates the neural network based on the loss score.

Shift invariant loss for deep learning based image segmentation

The present disclosure relates to systems and methods for cellular imaging and identification through the use of a light sheet flow cytometer. In one implementation, a light sheet flow cytometer may include a light source configured to emit light having one or more wavelengths, at least one optical element configured to form a light sheet from the emitted light, a microfluidic channel configured to hold a sample, and an imaging device. The imaging device may be adapted to forming 3-D images of the sample such that identification tags attached to the sample are visible.

Light sheet imaging flow cytometer

Systems and methods for generating virtually stained images of unstained samples are provided. According to an aspect of the invention, a method includes accessing an image training dataset including a plurality of image pairs. Each image pair includes a first image of an unstained first tissue sample, and a second image acquired when the first tissue sample is stained. The method also includes accessing a set of parameters for an artificial neural network, wherein the set of parameters includes weights associated with artificial neurons within the artificial neural network; training the artificial neural network by using the image training dataset and the set of parameters to adjust the weights; accessing a third image of a second tissue sample that is unstained; using the trained artificial neural network to generate a virtually stained image of the second tissue sample from the third image; and outputting the virtually stained image.

Generating virtually stained images of unstained samples

Systems and methods for fluorescent microscopy are disclosed where fluorophores can be excited over an excitation band to emit light in a wide emission band. Simultaneous acquisition of multiple planes in the sample can be achieved using a modified form of confocal microscopy. In one implementation, an objective employs a lens having optics exhibiting a large degree of axial chromatic aberration, such that emissions from different axially spaced focal planes are encoded by wavelength. Advantageously, simultaneous acquisition of multiple focal planes encoded by color can be processed to obtain efficient and rapid three-dimensional imaging of a sample.

Systems and methods for simultaneous acquisition of multiple planes with one or more chromatic lenses

Systems and methods for hyperspectral imaging are described. In one implementation, a hyperspectral imaging system includes a sample holder configured to hold a sample, an illumination system, and a detection system. The illumination system includes a light source configured to emit excitation light having one or more wavelengths and a diffractive element. The illumination system is configured to structure the excitation light into a predetermined two-dimensional pattern at a conjugate plane of a focal plane in the sample, spectrally disperse the structured excitation light in a first lateral direction, and illuminate the sample in an excitation pattern with the one or more wavelengths dispersed in the first lateral direction.

Systems and methods for 4-D hyperspectral imaging

Systems and methods for confocal imaging are described. In one implementation, a confocal imaging system may include a light source configured to emit excitation light having one or more wavelengths, a sample holder configured to hold a sample, a two-dimensional (2-D) imaging device, a first set of optical elements, and a second set of optical elements. The first set of optical elements may include a first spatial light modulator (SLM) and at least one lens. The first set of optical elements may together be configured to collimate the excitation light, apply a predetermined phase modulation pattern to the collimated excitation light, and illuminate the sample in an excitation pattern.

Spatial light modulator based hyperspectral confocal microscopes and methods of use

Systems and methods for dispersing an optical beam are disclosed. In one implementation, an optical system includes a first double Amici prism and a second double Amici prism. The first and second double Amici prisms are aligned along an optical axis of the system and configured to transmit the optical beam. At least one of the first and second double Amici prisms is rotatable relative to the other around the optical axis. Advantageously, the disclosed systems and methods allow for efficient and versatile adjustment of the magnitude and/or orientation of the dispersion of the optical beam.

Rotatable prisms for controlling dispersion magnitude and orientation and methods of use

Methods are provided to identify spatially and spectrally multiplexed probes in a biological environment. Such probes are identified by the ordering and color of fluorophores of the probes. The devices and methods provided facilitate determination of the locations and colors of such fluorophores, such that a probe can be identified. In some embodiments, probes are identified by applying light from a target environment to a spatial light modulator that can be used to control the direction and magnitude of chromatic dispersion of the detected light; multiple images of the target, corresponding to multiple different spatial light modulator settings, can be deconvolved and used to determine the colors and locations of fluorophores. In some embodiments, light from a region of the target can be simultaneously imaged spatially and spectrally. Correlations between the spatial and spectral images over time can be used to determine the color of fluorophores in the target.

Growing community of inventors

San Bruno, CA, United States of America

Cheng-Hsun Wu