Methods of generating trophectoderm and neurectoderm from human embryonic stem cells
Methods for inducing differentiation of stem cells into trophectoderm, neurectoderm, or progeny cells thereof are described, and mathematical and statistical image analysis methods and systems, for example, to identify differentiated cells and screen for agents which modulate differentiation are provided for separate or combined use.
1. A method for identifying cells differentiating into trophectoderm, neurectoderm, or progeny cells thereof, comprising:
(a) obtaining an image of one or more cells;
(b) representing the image as a multiplicity of pixels;
(c) using a processor to extract one or more image features from said multiplicity of pixels; and
(d) identifying stem cells differentiating into trophectoderm, neurectoderm, or progeny cells thereof by comparing the one or more image features with image features derived from one or more reference stem cells, wherein the processor applies one or more statistical comparison methods to determine a presence of discrete objects within cell borders of the one or more cells, as compared with absence of discrete objects, within the cell borders of the one or more reference stem cells.
2. The method of claim 1 , wherein high, low, or varying degrees of contrast between pixels at the cell border of a cell relative to the cell border of a reference stem cell indicates that the cell is differentiating into trophectoderm, neurectoderm, or a progeny cell thereof.
3. The method of claim 1 , wherein thin edges at borders of a cell, as compared with the edges of a cell border in a reference stem cell, indicates that the cell is differentiating into trophectoderm, neurectoderm, or a progeny cell thereof.
4. The method of claim 1 , wherein the content in multiple spatial frequency bands in an image of a cell relative to content in the multiple spatial frequency bands in reference image of a stem cell indicates that the cell is differentiating into trophectoderm, neurectoderm, or a progeny cell thereof.
5. The method of claim 1 , further comprising:
(e) performing simultaneous smoothing and segmentation of said image;
(f) determining one or more boundaries of said cell;
(g) identifying a region or subregion near to said one or more boundaries;
(h) deriving one or more attributes of region or subregion; and
(i) analyzing variation in said one or more attributes; wherein said one or more image features comprise components of said one or more attributes.
6. The method of claim 1 , wherein the one or more image features are extracted using a wavelet, or any multiresolution decomposition algorithm.
7. The method of claim 1 , wherein the one or more image features are extracted using a matrix edge and/or onion peel algorithms.
8. The method of claim 6 , wherein the wavelet or multiresolution decomposition algorithm is an n-level decomposition that yields three detail subbands per level.
9. The method of claim 1 , wherein the one or more statistical methods is a comparison of probability density functions, or estimates thereof.
10. A method for identifying cells differentiating into trophectoderm, neurectoderm, or progeny cells thereof, comprising:
(a) obtaining an image of one or more cells;
(b) representing the image as a multiplicity of pixels;
(c) using a processor to extract one or more image features from said multiplicity of pixels; and
(d) identifying cells differentiating into trophectoderm, neurectoderm, or progeny cells thereof by comparing the one or more image features with image features derived from one or more reference cells, wherein to compare the one or more image features the processor applies one or more statistical comparison methods to determine a presence of discrete objects within cell borders of the one or more cells, as compared with absence of discrete objects, within the cell borders of the one or more reference stem cells,
wherein the one or more statistical comparison methods include a parametric or non-parametric binary or M-ary hypothesis test and the one or more areas of the one or more images are classified using a clustering algorithm, wherein the clustering algorithm is further configured such that the dissimilarity between the one or more image features is calculated using a probability density function (pdf) estimator and quantified using information divergence.
11. The method of claim 10 , wherein the method used to estimate the pdf and divergence is selected from: a generalized Gaussian density model (GGD), a Gaussian density function, a non-Gaussian density function, or a non-parametric histogram.
12. The method of claim 10 , further comprising subdividing the image of one or more cells into one or more windows.
13. The method of claim 12 , wherein the one or more windows are classified, subdivided, and re-classified.
14. The method of claim 10 , wherein the appearance of squamous epithelium, as compared to the cuboidal appearance of reference stem cells, indicates that the cell is differentiating along the trophectoderm lineage, and the appearance of multinucleated cells indicates that the cells are differentiating along the trophectoderm lineage.