Astronomical Medicine & the Future of High-Dimensional Data Visualization and Analysis Software

Alyssa A. Goodman

Director of the Center for Innovative Computing, Astrophysics, Harvard University

 

Historically, astronomical data came in two flavors: images and spectra. Images showed two-dimensional maps of intensity, whether at optical or another wavelength. Spectra showed intensity as a function of wavelength at a particular position on the sky. Today, though, it is becoming increasingly common for data to take the form of "spectral-line maps," where "images" have a third dimension, usually wavelength. In the COMPLETE Survey of Star-Forming Regions (cfa-www.harvard.edu/COMPLETE), spectra with a thousand channels of wavelength information are acquired at hundreds of thousands of positions (pixels), creating data sets with of order >10^8 interesting voxels worth of information. To meet the challenge of understanding this kind of rich but massive data set, we have adapted software developed for medical image display and analysis for use on astronomical data. The software is based on the "VTK" toolkit, and builds on the programs known as "3D-Slicer" and "OsiriX." In the talk, I will explain why this kind of visualization and analysis software is becoming more and more crucial in the analysis of large astronomical, geophysical, and other survey projects. I will also outline our work on creating a generalized specification set for the multi-dimensional display and analysis tools of the future. The "Astronomical Medicine" project is one of a suite of startup projects at Harvard's new Initiative in Innovative Computing (iic.harvard.edu), and I will conclude the talk with a short discussion of the IIC's general goals and current activities.