A Kolmogorov metric embedding for live cell microscopy signaling patterns.

Summary: We present a metric embedding that captures spatiotemporal patterns of cell signaling dynamics in 5D (x, y, z, channel , time ) live cell microscopy movies. The embedding uses a metric distance called the normalized information distance (NID) based on Kolmogorov complexity theory, an absolute measure of information content between digital objects. The NID uses statistics of lossless compression to compute a theoretically optimal metric distance between pairs of 5D movies, requiring no a priori knowledge of expected pattern dynamics, and no training data. The cell signaling structure function (SSF) is defined using a class of metric 3D image filters that compute at each spatiotemporal cell centroid the voxel intensity configuration of the nucleus w.r.t. the surrounding cytoplasm, or a functional output, e.g. velocity. The only parameter is the expected cell radii ( μ m ). The SSF can be optionally combined with segmentation and tracking algorithms. The resulting lossless compression pipeline represents each 5-D input movie as a single point in a metric embedding space. The utility of a metric embedding follows from Euclidean distance between any points in the embedding space approximating optimally the pattern difference, as measured by the NID, between corresponding pairs of 5D movies. This is true throughout the embedding space, not only at points corresponding to input images. Examples are shown for synthetic data, for 2D+time movies of ERK and AKT signaling under different oncogenic mutations in human epithelial (MCF10A) cells, for 3D MCF10A spheroids under optogenetic manipulation of ERK, and for ERK dynamics during colony differentiation in human induced pluripotent stem cells. Availability and implementation: All of the software, including the phantom data generation and analysis, is available free and open-source, as described in the 'Data Availability' section.