AnatLab: Virtual Anatomy Laboratory

Notes on a talk presented by Michael Doyle and Cyndy Lilagan at the Tcl2008 conference.

AnatLab: "Google Earth for the human body."

• Web multidimensional atlas of human anatomical information
• Tcl-based anatomical nameserver
• zMap-enhanced tclhttpd Web server
• Javascript/Ajax-capable and Flash-capable Web browser can explored the Visible Human male dataset
• tested on Safari, Firefox and IE
• currently in beta at the University of Illinois College of Medicine

(... showed medical anatomy images from cadavers! ...)

Features:

• slice through data--query by clicking objects
• displays name and highlights structure
• user can zoom from overview scaled images into detailed higher magnification views
• "more..." links opens up a separate window and displays corresponding Wikipedia content
• as user clicks on other structures the system automatically updates the Wikipedia window

Tcl dogfood used to develop AnatLab:

• Groupware: Mark Roseman's ProjectForum
• Code Repository: DRH's Fossil
• TclHTTPd

Image processing: Img used on anatomical object data (~2500 objects) and imagemaps (1878 maps) to populate a sqlite database with object names and object section ranges, and to create section images with highlighted objects (tkimgwin.kit also used).

Summer '08 Annotation Team: anatomist, neuro-anatomist, 4 medical students, quality control assistants and a programmer. QC assistance used SQLite-adapted "A little image viewer" (by RS) to look at highlighted section images to visually compare between the AnatLab and the National Library of Medicine's "Atlas of the Visible Human Male" printed atlas. Medical students took Mike Doyle's tcl-based annotator program, anatomists' directions, a QC comparison and their medical education background to add approximately 150 new objects and 6,700 new annotations to the current atlas. [showed example of the annotation tool and how it was used] These annotations were used to create a new imagemaps and new highlighted section images for feedback and corrections. The special case of preserving old objects within newly annotated objects was also covered. SEH replaced the original Java mapping tool with a Tcl version (and it runs faster).

Future directions:

• A web-based collaborative environment using tcl-plugin or Aejaks.
• Combination of annotation creation and annotation processing for immediate feedback and corrections.
• Specific immediate enhancements:

1. Annotations of very small objects using a magnify facility (Keith Vetter's magnifying glass[L1 ])
2. Use of the sqlite database for object identification.
3. Use of object traced images (150K) to generate polygon annotations.

Plus various long-term enhancements and bug fixes.

(... Mike Doyle ran a live demo ...)