Apparatus, method, and system for automated, non-invasive cell activity tracking

1. A method for automated, non-invasive evaluation of cell activity in human embryos, oocytes, or pluripotent cells to determine a developmental potential with an imaging system, the method comprising:

acquiring a series of time-sequential images of the cells in a multi-well culture dish with a camera of at least one time-lapse microscope of the imaging system;

selecting an image from the series of time-sequential images, wherein the selected image is preceded by a previous image in the series of time-sequential images of the cells;

extracting observable geometric information of the cells from the selected image;

generating a plurality of first hypotheses characterizing the cells as shown in the image, wherein generating the plurality of first hypotheses comprises:

determining an inferred characteristic of one or more of the cells based on a mapping of a representation of each of the one or more of the cells to the observable geometric information associated with the cells;

retrieving a plurality of parent hypotheses associated with the previous image in the series of time-sequential images of the cells;

generating a plurality of preliminary hypotheses based on the plurality of parent hypotheses; and

generating the plurality of first hypotheses based on the plurality of preliminary hypotheses and based on the observable geometric information of the cells from the selected image;

selecting a first hypothesis from the plurality of first hypotheses associated with the image; and

determining a characteristic of the one or more of the cells based on the inferred characteristic associated with the first hypothesis to determine an indicator of developmental potential for the cells.

2. The method of claim 1 , further comprising:

generating a preliminary hypothesis characterizing the cells as shown in the image; and

refining the preliminary hypothesis to obtain one or more of the plurality of first hypotheses based on the mapping.

3. The method of claim 1 , wherein the observable geometric information comprises at least one of:

a plurality of cell boundary points;

a plurality of cell boundary segments;

a shape of the cells in the multi-well culture dish; and

an arrangement of the cells in the multi-well culture dish.

4. The method of claim 3 , wherein the inferred characteristic of one or more of the cells is determined based on explicit mapping.

5. The method of claim 4 , wherein the inferred characteristic of the one or more cells can be at least one of: an inferred geometry of the one or more cells; and an inferred number of the one or more cells.

6. The method of claim 1 , wherein generating a plurality of preliminary hypotheses based on the plurality of parent hypotheses further comprises:

sampling at least one aspect of the parent hypotheses; and

perturbing the at least one aspect of the parent hypotheses.

7. The method of claim 6 , wherein the at least one aspect of the parent hypotheses comprises a plurality of ellipses.

8. The method of claim 1 , wherein the plurality of first hypotheses are generated through expectation maximization (EM) optimization.

9. The method of claim 1 , wherein the plurality of first hypotheses are generated by:

generating a mapping representation of cells to boundary segments;

refining the preliminary hypotheses based on the mapping to obtain refined hypotheses; and

scoring the refined hypotheses based on the observable geometric information.

10. The method of claim 9 , further comprising:

selecting a second, subsequent image from the series of time-sequential images;

generating a plurality of second hypotheses characterizing the cells as shown in the second, subsequent image;

and selecting a second hypothesis from the plurality of second hypotheses,

wherein the first hypothesis is selected from the plurality of first hypotheses according to a first approximate inference over a probabilistic graphical model, and wherein the second hypothesis is selected from the plurality of second hypotheses according to a second approximate inference over the probabilistic graphical model.

11. An automated imaging system for evaluation of human embryos, oocytes, or pluripotent cells to determine a developmental potential, the system comprising:

a stage configured to receive a multi-well culture dish, wherein the multi-well culture dish comprises a plurality of micro-wells containing a plurality of human embryo cells or pluripotent cells; and

a time-lapse microscope having at least a camera, wherein the camera is configured to acquire a series of time-sequential images of the multi-well culture dish on the stage, and wherein the time-lapse microscope is configured to determine an indicator of developmental potential of at least some of the cells, wherein determining the indicator of development competence comprises:

selecting an image from the series of time-sequential images, wherein the selected image is preceded by a previous image in the series of time-sequential images of the cells;

extracting observable geometric information of the cells from the selected image;

generating a plurality of first hypotheses characterizing the cells as shown in the image, wherein generating the plurality of first hypotheses comprises:

determining an inferred characteristic of one or more of the cells based on a mapping of a representation of each of the one or more of the cells to the observable geometric information associated with the cells;

retrieving a plurality of parent hypotheses associated with the previous image in the series of time-sequential images of the cells;

generating a plurality of preliminary hypotheses based on the plurality of parent hypotheses; and

generating the plurality of first hypotheses based on the plurality of preliminary hypotheses and based on the observable geometric information of the cells from the selected image;

selecting a first hypothesis from the plurality of first hypotheses associated with the image; and

determining a characteristic of the one or more of the cells based on the inferred characteristic associated with the first hypothesis.

12. The automated imaging system of claim 11 , wherein the observable geometric information comprises at least one of:

a plurality of cell boundary points;

a plurality of cell boundary segments;

a shape of the cells in the multi-well culture dish; and

an arrangement of the cells in the multi-well culture dish.

13. The automated imaging system of claim 12 , wherein the inferred characteristic of one or more of the cells is determined based on explicit mapping.

14. The automated imaging system of claim 13 , wherein the inferred characteristic of the one or more cells can be at least one of: an inferred geometry of the one or more cells; and an inferred number of the one or more cells.

15. The automated imaging system of claim 11 , wherein generating a plurality of preliminary hypotheses based on the plurality of parent hypotheses further comprises:

sampling at least one aspect of the parent hypotheses; and

perturbing the at least one aspect of the parent hypotheses.

16. The automated imaging system of claim 15 , wherein the at least one aspect of the parent hypotheses comprises a plurality of ellipses.

17. The automated imaging system of claim 11 , wherein the plurality of first hypotheses are generated through expectation maximization (EM) optimization.

18. The automated imaging system of claim 11 , wherein the plurality of first hypotheses are generated by:

generating a mapping representation of cells to boundary segments;

refining the preliminary hypotheses based on the mapping to obtain refined hypotheses; and

scoring the refined hypotheses based on the observable geometric information.

19. The automated imaging system of claim 18 , wherein the time-lapse microscope is further configured to:

select a second, subsequent image from the series of time-sequential images;

generate a plurality of second hypotheses characterizing the cells as shown in the second, subsequent image; and

select a second hypothesis from the plurality of second hypotheses,

wherein the first hypothesis is selected from the plurality of first hypotheses according to a first approximate inference over a probabilistic graphical model, and wherein the second hypothesis is selected from the plurality of second hypotheses according to a second approximate inference over the probabilistic graphical model.