Autonomous cell imaging and modeling system
Inventors
Marie-Nelly, Hervé • VELAYUTHAM, Jeevaa • Phillips, Zachary • TU, Shengjiang
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12198344-B2
Publication Date
2025-01-14
Expiration Date
Abstract
The present disclosure relates generally to an autonomous cell imaging and modeling platform, and more specifically to machine-learning techniques for using microscopy imaging data to continuously study live biological cells. The autonomous cell imaging and modeling platform can be applied to evaluate various cellular processes, such as cellular differentiation, optimization of cell culture (e.g., in-plate cytometry), disease modeling, histopathology imaging, and genetic and chemical screening, using a dynamic universal imaging system. In some embodiments, the platform comprises a set of label-free computational imaging techniques, self-supervised learning models, and robotic devices configured in an autonomous imaging system to study positional and morphological characteristics in particular cellular substructures of a cell culture in an efficient and non-destructive manner over time.
Core Innovation
The invention relates to a system for modeling a characteristic of interest of a cell culture comprising one or more live mammalian cells. The system obtains a first set of one or more images capturing the cell culture at a first time point and inputs the first set of images into a trained machine-learning model to obtain a first set of embedding vectors representing positional and morphological characteristics in particular cellular substructures. The system evaluates the characteristic of interest corresponding to the first time point based on the first set of embedding vectors to obtain a first set of one or more values.
The system obtains a second set of one or more images capturing the cell culture at a second time point and inputs the second set into the trained machine-learning model to obtain a second set of embedding vectors representing positional and morphological characteristics in the particular cellular substructures. The system evaluates the characteristic of interest corresponding to the second time point based on the second set of embedding vectors to obtain a second set of one or more values and determines, from the first set of values and the second set of values, a change of the characteristic of interest in the cell culture.
The approach is described as an autonomous, label-free cell imaging and modeling platform that repeatedly images live mammalian cell cultures over time, including quantitative phase imaging and optional fluorescence/autofluorescence via modality transformation using GANs, and uses self-supervised learning to produce dynamic embedding vectors for downstream modeling and assessment.
Claims Coverage
The provided relevant claims include three independent claims. Each claim centers on time-point image acquisition of live mammalian cells, conversion of images into embedding vectors by a trained machine-learning model, evaluation of a characteristic of interest from the embeddings, and determining a change in the characteristic between two time points.
Two-timepoint embedding-based modeling of a cell culture characteristic
Obtain a first set of one or more images capturing the cell culture at a first time point, input the first set into a trained machine-learning model to obtain a first set of embedding vectors representing positional and morphological characteristics in particular cellular substructures, evaluate the characteristic of interest corresponding to the first time point based on the first set of embedding vectors to obtain a first set of one or more values, obtain a second set of images at a second time point, input the second set into the trained machine-learning model to obtain a second set of embedding vectors, evaluate the characteristic of interest corresponding to the second time point based on the second set of embedding vectors to obtain a second set of one or more values, and determine from the first set of values and the second set of values a change of the characteristic of interest in the cell culture.
Non-transitory program storage for embedding-vector evaluation of change
Store one or more programs comprising instructions that, when executed by one or more processors of an electronic device, cause operations of obtaining images at a first time point and a second time point, inputting each image set into a trained machine-learning model to obtain embedding vectors representing positional and morphological characteristics in particular cellular substructures, evaluating the characteristic of interest at each time point based on the embedding vectors to obtain corresponding values, and determining a change of the characteristic of interest based on the first set of values and the second set of values.
Method of modeling characteristic change using embedding vectors from live-cell images
A method comprising obtaining a first set of one or more images capturing the cell culture at a first time point, inputting the first set into a trained machine-learning model to obtain a first set of embedding vectors representing positional and morphological characteristics in particular cellular substructures, evaluating the characteristic of interest corresponding to the first time point based on the first set of embedding vectors to obtain a first set of one or more values, obtaining a second set of images capturing the cell culture at a second time point, inputting the second set into the trained machine-learning model to obtain a second set of embedding vectors, evaluating the characteristic of interest corresponding to the second time point based on the second set of embedding vectors to obtain a second set of one or more values, and determining from the first set of one or more values and the second set of one or more values a change of the characteristic of interest in the cell culture.
Across the independent claims, the core claimed coverage is an embedding-vector pipeline tied to two time points: live mammalian cell imaging, embedding generation by a trained machine-learning model, embedding-based evaluation of a characteristic of interest, and determination of a change in that characteristic between the first and second time points.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?