The movies that we work with are time lapse image sequences of live organelles, cells and tissue. Here are a few examples. For more details, the publications page has publicly accessible links to most of the papers that are referenced. For best results – play the movies at 1080p in full screen mode.

This movie shows the analysis of mouse adult neural stem cells using our segmentation, tracking and lineaging algorithms. The segmentation includes both nucleus and processes (left), with the bounding box indicating the process extent and the nuclear location along the processes. The lineage tree (right) shows cell fate commitment using retrospective immunohistochemistry. Normalized nucleus and process size are also plotted on the lineage tree. Note how progenitors that will produce different cell types have such clear differences in process size and dynamics…

This is a supplementary movie from:

E. Cardenas De La Hoz, M. R. Winter, M. Apostolopoulou, S. Temple, A. R. Cohen, “Measuring Process Dynamics and Nuclear Migration for Clones of Neural Progenitor Cells”,  ECCV 2016 Workshops, Volume 9913, Springer Lecture Notes in Computer Science pp 291-305.

Here is a movie showing the development of a clone of neural stem cells. This movie was made by the new version of our LEVER program. We segment each cell in each image frame. We then track the cells, figuring out which segmentation is which in each frame. Finally, we generate the lineage tree. A human validates all of this, correcting any errors as necessary. This movie has 303 cells, and 30,453 segmentations. 1.8% of the automated results had to be corrected by hand – this is a big improvement over existing options, and we already know how to do better! More about LEVER here…

I think this is a cool movie, but that movies are a terrible way to look at this data! Instead, you should click here to use our new tool, CloneView to look at this movie. All of the image data that was used in the paper can be seen here:
Reference: M. Winter, M. Liu, D. Monteleone, J. Melunis, U. Hershberg, S. K. Goderie, S. Temple, and A. R. Cohen, Computational Image Analysis Reveals Intrinsic Multigenerational Differences Between Anterior and Posterior Cerebral Cortex Neural Progenitor Cells, Stem Cell Reports. 2015. In Press,

Here’s another movie from the same paper as above. This shows how the LEVER program uses occasional manual user edits to automtically correct the vast majority of the segmentation and tracking errors.

If you look closely you can see tracking information on the lineage tree (here in the left panel) change as tracking errors are automatically updated in response to the improved segmentation. This is our ‘segmentation refinement’ algorithm in action!

This next movie is actually not a time lapse image sequence. It is a static image with no time component. This movie shows a montage composed of 74 overlapping 3-D image stacks. The image resolution is 10,173 x 3,858 x 74 voxels! Each image stack contains 5 channels. We zoom in from full resolution, cutting the field of view in half with each zoom, until we reach the final unscaled resolution. The channels are: blood vessels (red), cell nuclei (dark blue), neural stem cells and astrocytes (green), oligodendrocytes (yellow), and migrating neuroblasts (cyan).

E. Wait, M. Winter, C. Bjornsson, E. Kokovay, Y. Wang, S. Goderie, S. Temple and A. R. Cohen, Visualization and Correction of Automated Segmentation, Tracking and Lineaging from 5-D Stem Cell Image Sequences. BMC Bioinformatics, 15:328. doi:10.1186/1471-2105-15-328. October 3, 2014.

This movie is a “5-D” image sequence. The first three dimensions are (x,y,z). Time is dimension number four. The fifth dimension is the microscope imaging channel, with e.g. blood vessels and ependyma on one channel and neural stem cells on another. Here we can see the relationship between the stem cells and the blood vessels that form their all-important “niche”. The cleavage plane drawn during mitosis (complete with Matrix spin effect) quantifies the “polarity” of the division, or the orientation of the daughter cells relative to the nearest vasculature.

E. Wait, M. Winter, C. Bjornsson, E. Kokovay, Y. Wang, S. Goderie, S. Temple and A. R. Cohen, Visualization and Correction of Automated Segmentation, Tracking and Lineaging from 5-D Stem Cell Image Sequences. BMC Bioinformatics, 15:328. doi:10.1186/1471-2105-15-328. October 3, 2014.

This movie shows mouse adult neural stem cells (NSCs) in the top panel, with embryonic NSCs in the bottom panel. The scale bars indicate 25µm. The lineage trees (right panel) show the development of the clone (family tree) from a single cell through terminal differentiation. Note the difference in size, motility and population dynamics.

Winter, M., E. Wait, B. Roysam, S. Goderie, E. Kokovay, S. Temple, and A.R. Cohen, Vertebrate Neural Stem Cell Segmentation, Tracking and Lineaging with Validation and Editing. Nature Protocols, 2011. 6(12): p. 1942-52. (PMC)

This movie shows organelle transport along the axon (right panel) together with a kymograph (left panel). We are tracking the organelles directly on the image, and showing results on the kymograph. Notice how the kymograph, like the lineage tree above, is an effective dimensionality reduction technique for the image sequence data. Our goal here is to measure changes in organelle transport in neurodegenerative vs. wild type populations.

Winter, M., C. Fang, G. Banker, B. Roysam, and A. R. Cohen, Axonal Transport Analysis Using Multitemporal Association Tracking. International Journal of Computational Biology and Drug Design 2012;5(1):35-48. (PMC)

In this final movie, cells 11 and 300 are rat retinal progenitor cells. Cell 300 will produce 1 photoreceptor neuron and 1 amacrine cell; cell 11 will produce 2 photoreceptor neurons. We have developed new segmentation (blue outlines) and tracking (red numbers) algorithms to extract the patterns of motion, morphology and inter-cellular assocation. Analyzing the segmentation and tracking results, our computational tools discovered behavioral differences among retinal progenitors that can accurately predict the type of daughter cells a retinal progenitor will produce, while the progenitor is alive inside the microscope.

A.R. Cohen, F. Gomes, B. Roysam, and M. Cayouette, Computational prediction of neural progenitor cell fates. Nature Methods, 2010. 7(3): p. 213 – 218.