Further Projects

These are interdisciplinary projects AM members are conducting with external collaborators.

Please see our general projects page for an overview.

3D Corneal Imaging

Most medical data is high contrast (e.g. muscle vs. bone), symmetric, and can be mapped to an "atlas" (i.e. every human has the same general anatomy). However, this is not true for all medical imaging. One such example is the human cornea which is the transparent outer layer of the human eye. Each human cornea looks slightly different, and certain diseases (such as keratoconus or fuchs' corneal dystrophy) will drastically change its appearance. In order to enhance diagnosis of these corneal pathologies, 3D images must be analyzed and reviewed by a specialist. Conventional data visualization and segmentation techniques do not work well for these data sets, and the typical approach has been to look at the raw imagery as 2D slices through the data cube. However, their inhomogeneous and "nebulous" appearance makes the data work well with astronomical techniques. We have been collaborating with ophthalmologists to better visualize their 3D corneal data sets generated by confocal microscope imaging with the goal of improving the diagnosis timeline and subsequent intervention. We are also working on further interdiciplanary projects which are beneficial to both medical and astronomy imagers.

Contributors: Michelle Borkin (IIC), Dr. Taha Ahmed (Tennent Institute of Ophthalmology, Glasgow, UK), Dr. Kanna Ramaesh (Tennent Institute of Ophthalmology, Glasgow, UK), & Michael Halle (SPL/IIC)

Segmentation of Astronomy data with EMSegment

In addition to the visualization of 2D/3D data, another important aspect common to both astronomy and medical imaging is the need for data segmentation. In the medical community, this would be applied to automatic labeling of anatomical structures in an MRI scan. In the astronomy community, this would mean picking out all the stars in your telescope image or the dense cores inside a nebula. We have been working to bridge these two communities by applying segmentation algorithms developed for specific fields to others. Our first step has been working with EMSegment, an Expectation-Maximization (EM) algorithm segmentation module in 3D Slicer originally developed for brain segmentation, and applying it to the detection of young stars and star forming cores. We plan to continue this research and expand in the future.

Contributors: Michael Halle (SPL/IIC), Kilian Pohl (BWH/HMS), Michelle Borkin (IIC), & Jens Kauffmann (IIC/CfA)

3D Stereo Imaging of Astronomy Data

Stereoscopic images have been used extensively in science throughout the years to help researchers understand the shape and structure of complex objects. As computers have become essential parts of the scientist's workbench, computer-generated stereo images of raw and processed data have become important for a variety of research fields. For example, geologists use 3D to help identify possible regions for new oil and gas reserves from earth-penetrating radar signals. The pharmaceutical industry, and most recently the field of genomics, employs stereo to see how drugs interact with organic molecules in our body. Stereo imaging provides another tool for data comprehension on top of what computer graphic reconstruction can offer to astronomers. Stereo “pulls apart” objects located as different depths in the data volume. It helps reveal partially obscured structures corresponding to objects inside the cloud. Perhaps most importantly, though, stereo images provide researchers with a better mental model about the complex shape and subtle structure of data features never before envisioned.

We have been utilizing different kinds of stereo display (e.g. slide pairs, 3D project technology, lenticular prints, etc.) in order to help astronomers better understand their 3D data sets, and see which methods prove the most useful. We are also in the process of collaborating with the Durham Visualization Lab on a sterographic educational movie about the early life of stars. The movie is driving both the development of new 3D movie editing and production techniques, as well as creating new 3D and stereo visualizations of star forming regions for astronomical research. If you are interested in viewing some of our stereo image pairs, you can download our pdf handout.

Contributors: Michelle Borkin (IIC), Nick Holliman (Durham University, UK), Michael Halle (SPL/IIC), Jens Kauffmann (IIC/CfA), & Alyssa Goodman (CfA)

Outflow Nebulosity in star-forming Regions

In strict terms, this is a mere astronomical research project, for which AM received NASA funding. The plan is to catalogue nebulous spots of gas that is ejected from young stars. This search is based on Spitzer Space Telescope infrared images collected by the Cores to Disks (c2d) Spitzer legacy project.

In a more general perspective, however, the expertise gained in this domain of star formation research feeds back into the core projects described elsewhere. This might in particular inform us which features are still not detected in radio telescope data even when using 3D visualization. Also, this gives us a first hand experience with extraction of complex shapes from images, a skill that might pay back in the future. This project allowed us to have a dedicated MSc student working with us, which helps shaping our educational profile.

Project Lead: Jens Kauffmann (IIC/CfA)

Contributors: Katherine Guenthner (Leipzig), Hector Arce (Yale), Karl Stapelfeldt (JPL), Neal Evans (U Texas), Lori Allen (CfA), Tyler Bourke (CfA), Tracy Huard (Maryland), Michelle Borkin (IIC), Jaime Pineda (CfA), Jonathan Foster (CfA) & Alyssa Goodman (CfA)