1 / 23

Finding correspondences between drosophila embryo

Finding correspondences between drosophila embryo. Charless Fowlkes, Jitendra Malik UCB Computer Sciences. ?. % egg length coordinate system. 10% strip. Warp x-axis to align peak features. Myasnikova et al (2001). Advantages: Treats cells as discrete units

walker
Download Presentation

Finding correspondences between drosophila embryo

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Finding correspondences between drosophila embryo Charless Fowlkes, Jitendra Malik UCB Computer Sciences

  2. ?

  3. % egg length coordinate system 10% strip Warp x-axis to align peak features Myasnikova et al (2001)

  4. Advantages: Treats cells as discrete units Discontinuities in expression levels between cells are preserved Given a correspondence, it is easy to find a global coordinate system Cell-to-Cell Correspondence • Cell lineage gives “ground-truth” correspondence

  5. Sources of Variation • Not so interesting: • Staining • Shrinking • Spinning • Squashing • Staging • Interesting: • Biological Variation

  6. Correspondence as Optimzation • Of all possible correspondences between two embryos, find the one which pairs cells with similar expression levels.

  7. Correspondence as Optimzation • Of all possible correspondences between two embryos, find the one which pairs cells with similar expression levels.

  8. Correspondence as Optimzation • Of all possible correspondences between two embryos, find the one which pairs cells with similar expression levels and results in a smooth transformation

  9. Correspondence Overview Nuclear Segmentation Coarse Registration X Y Z Sna Ftz 220.3 40.2 31.0 0.28 0.99 182.4 30.1 44.9 0.22 0.77 . . Detailed Correspondence and Model Building Mesh Generation Feature Extraction

  10. Coarse Registration • Align the center of masses of two embryos • Scale along AP axis in order to match length • Sphere perpendicular to AP axis in order to remove squashing • Rotate along AP axis to align SNA expression pattern

  11. Coarse Registration Scale and Rotate

  12. Coarse Consensus Model

  13. Correspondence Overview Nuclear Segmentation Coarse Registration X Y Z Sna Ftz 220.3 40.2 31.0 0.28 0.99 182.4 30.1 44.9 0.22 0.77 . . Detailed Correspondence and Model Building Mesh Generation Feature Extraction

  14. Mesh generation • Point cloud doesn’t capture the blastoderm topology. Locally, it is a 2D sheet of cells

  15. Mesh generation • Utilize standard tools from computer graphics and computational geometry

  16. Feedback Mesh Information to Segmentation ? Nuclear Segmentation X Y Z Sna Ftz 220.3 40.2 31.0 0.28 0.99 182.4 30.1 44.9 0.22 0.77 . . Mesh Generation

  17. Feedback Mesh Information to Segmentation ?

  18. Correspondence Overview Nuclear Segmentation Coarse Registration X Y Z Sna Ftz 220.3 40.2 31.0 0.28 0.99 182.4 30.1 44.9 0.22 0.77 . . Detailed Correspondence and Model Building Mesh Generation Feature Extraction

  19. Feature Extraction • Extract expression levels in the neighborhood of a given cell by flattening the sheet.

  20. Correspondence Overview Nuclear Segmentation Coarse Registration X Y Z Sna Ftz 220.3 40.2 31.0 0.28 0.99 182.4 30.1 44.9 0.22 0.77 . . Detailed Correspondence and Model Building Mesh Generation Feature Extraction

  21. Detailed Correspondence • Set up quadratic optimzation problem which includes matching and smoothness terms • only considers correspondences between points which are nearby after the coarse registration • 100 fold reduction in potential matches • utilize features extracted from the mesh • much better match accuracy • further pruning • Results…soon

  22. Conclusion • Techniques aimed at cellular correspondence between multiple embroys • Utilize knowledge of embryo topology along with rough alignment to make detailed correspondence tractable

More Related