Michael J. Ackerman, Ph.D.
Office of High Performance Computing and Communications
National Library of Medicine, Bethesda, MD
D-Lib Magazine, October 1995
The Visible Human Project has its roots in a 1986 long-range planning effort of the National Library of Medicine (NLM). It foresaw a coming era where NLM's bibliographic and factual database services would be complemented by libraries of digital images, distributed over high speed computer networks and by high capacity physical media. Not surprisingly, it saw an increasing role for electronically represented images in clinical medicine and biomedical research. It encouraged the NLM to consider building and disseminating medical image libraries much the same way it acquires, indexes, and provides access to the biomedical literature. As a result of the deliberations of consultants in medical education, the long-range plan recommended that the NLM should "thoroughly and systematically investigate the technical requirements for and feasibility of instituting a biomedical images library."
Early in 1989, under the direction of the Board of Regents, an ad hoc planning panel was convened to explore in-depth the proper role for the NLM in the rapidly changing field of electronic imaging. After much deliberation, the NLM Planning Panel on Electronic Image Libraries made the following recommendation: "NLM should undertake a first project building a digital image library of volumetric data representing a complete, normal adult male and female. This Visible Human Project will include digitized photographic images for cryosectioning, digital images derived from computerized tomography and digital magnetic resonance images of cadavers."
The initial aim of the Visible Human Project is the acquisition of transverse CT, MRI and cryosection images of a representative male and female cadaver at an average of one millimeter intervals. The corresponding transverse sections in each of the three modalities are to be registered with one another. A contract for acquisition of these pixel-based data was awarded in August 1991 to the University of Colorado at Denver. Victor M. Spitzer, Ph.D. and David G. Whitlock, MD, Ph.D. are the principal investigators.
It took almost two years to find an appropriate male cadaver and two and a half years to find an appropriate female cadaver. To date only the male data set is available. It consists of Magnetic Resonance Interferometry (MRI), Computerized Tomography (CT) and anatomical images. Axial MRI images of the head and neck and longitudinal sections of the rest of the body were obtained at 4 mm intervals. The MRI images are 256 pixel by 256 pixel resolution. Each pixel has 12 bits of grey tone resolution. The CT data consists of axial CT scans of the entire body taken at 1 mm intervals at a resolution of 512 pixels by 512 pixels where each pixel is made up of 12 bits of grey tone. The axial anatomical images are 2048 pixels by 1216 pixels where each pixel is defined by 24 bits of color, about 7.5 megabytes. The anatomical cross-sections are also at 1 mm intervals and coincide with the CT axial images. There are 1871 cross-sections for each mode, CT and anatomy. The complete male data set is 15 gigabytes in size.
Sample full scale images are available via NLM's FTP site: nlmpubs.nlm.nih.gov. Six full color anatomical images and an explanatory README file can be found on the FTP site in visible/bitmaps/color24 as *.raw. Please be careful as each of these images is over 7 megabytes in size. Ten CT scan images and an explanatory README file can be found in visible/bitmaps/ct as *.fre (5 images captured while the cadaver was fresh) and *.fro (5 images captured after the cadaver was frozen). Six MRI scan images and an explanatory README file can be found in visible/bitmaps/mri as *.t1.
Scaled down versions of all of these image files can be found on NLM's FTP site in visible/gifs as *.gif and on the NLM Gopher, gopher.nlm.nih.gov, in the Visible Human Project section. On the world wide web, http://www.nlm.nih.gov, the images can be found in the Visible Human Project section as *.jpg.
A License Agreement for use of Visible Human Project data sets is required. It can be retrieved from NLM's gopher site. The Agreement will be found in the section entitled Visible Human Project as a text file and as a downloadable WordPerfect file. It is also available from NLM's FTP site. The Agreement will be found in section "visible" as a WordPerfect file, vhpagree.wp, or as a text file, vhpagree.txt. Two original signed copies of the agreement are required as well as a statement that explains your intended use of the data set. Send both signed copies of the Agreement and the "use" statement to NLM's Visible Human Project office. The agreement will be signed by NLM and one of the originals returned. Your account and password to the Visible Human Project Internet FTP site will also be returned as well as information on where you may purchase the data set on 8 mm Exabyte or 4 mm DAT tape. There are 6 tapes of anatomical images corresponding to 6 body regions: head, thorax, abdomen, pelvis, thighs, and feet. A 7th tape contains all the MRI and CT images. Each tape costs $150 in the US, Canada and Mexico, $300 elsewhere. An 8th, "sample", tape which contains anatomical images from the entire body at 1 cm increments is also available. The set of 8 tapes costs $1,000 in the US, Canada and Mexico, $2,000 elsewhere.
The data set is being distributed from the FTP site and on tape in a UNIX TAR format. Each file is individually compressed in a UNIX-Z compression format. UNIX systems should have a utility to do the decompression. The decompression may be accomplished on DOS based systems using a program called COMP430D. MacCompress 3.2 can be used for this purpose on Macintosh systems. Both programs are available from NLM's anonymous FTP site, nlmpubs.nlm.nih.gov, or from the Visible Human master FTP site, to do the decompression.
Each file on the tape or on the FTP site contains only one image. The anatomy and CT files are numbered from 1001 to 2878. If you subtract 1000 from the file number it will tell you the location of the file in millimeters from the top of the head. CT, anatomy and MRI file names which contain the same number are corresponding files. If a file number is missing then the cross-section at that level is not available. MRI cross- sectional images are only available for the head and neck and therefore corresponding numbers will only be found in that region.
The easiest way to explain how to read each of the data types is to explain how they would be read if you were using Photoshop 3.0. Under RAW input format, the values from the following table would be entered:
This will allow you to see the anatomy images. The CT and MRI images will appear as a black screen. For CT and MRI, set the "Mode" to "Greyscale". Then set "Image" to "Adjust" to "Auto". The CT and MRI images will now be visible.
Some users have asked if there are any minimum specifications for a computer in order to use the Visible Human Data Set. Unfortunately, there are none. Some people are using powerMacs and high level PC's. Others are using UNIX workstations. It all depends on what you want to do with the data. For one user who wanted to visually move around inside the data set in real time, even a Reality Engine was not big enough. For school children who want to see less detailed cross-sections, a 386 based computer is sufficient. Just remember each original anatomic image is 7 mbytes and there are 1871 of them. A user workstation needs to be scaled based on the image size, what you expect to do with that image, and how many images need to manipulated at one time.
The data set from the female cadaver will have the same characteristics as the male cadaver with one exception. The axial anatomical images will be obtained at 0.33 mm intervals instead of 1.0 mm intervals. This will result in over 5,000 anatomical images. The data set is expected to be about 40 gigabytes in size. Distribution is anticipated during the late fall of 1995. The spacing in the "Z" direction will be reduced to 0.33 mm in order to match the 0.33 mm pixel spacing in the "XY" plane. This will enable developers who are interested in three-dimensional reconstructions to work with cubic voxels.
The larger, long-term goal of the Visible Human Project is to produce a system of knowledge structures, which will transparently link visual knowledge forms to symbolic knowledge formats. Methods need to be developed to link image data to symbolic text-based data which is comprised of names, hierarchies, principles and theories. Although experiments are being done using generalizable linkage methods, like the use of hypermedia where words can be used to find pictures, and pictures can be used as an index into relevant text, standards do not currently exist for such linkages. Basic research also is needed in the description and representation of structures, and the connection of structural-anatomical to functional-physiological knowledge. Some of these issues are now being address in the development of extensions to the Metathesaurus under NLM's Unified Medical Language System (UMLS) program. The goal is to make the print library and the image library a single, unified resource for health information.
For additional information on the Visible Human Project please contact:
Michael J. Ackerman, Ph.D.
Visible Human Project
National Library of Medicine
8600 Rockville Pike
Bethesda, MD 20894
FAX: (301) 402-4080
National Library of Medicine (US) Board of Regents. "Electronic Imaging: Report of the Board of Regents". US Department of Health and Human Services, Public Health Service, National Institutes of Health, 1990. NIH Publication 90-2197. Available from: Office of Public Information, National Library of Medicine, 8600 Rockville Pike, Bethesda, Maryland 20894. (Phone: (301) 496-6308; Toll free 1-800-272-4787. Select codes 1,1,3,1. Fax: (301) 496-4450. E-mail address, Internet: email@example.com
(The figure in the table that refers to Anatomy Width has been corrected from 2040 to 2048 at the request of the author. Change made 2/14/00.)