|
| |
Visible Human Project
From: University of Michigan
| By:
Brian D. Athey |
EDITOR'S INTRODUCTION |
The University of Michigan's Visible Human Project is creating detailed three-dimensional image databases of the human body--a virtual anatomy atlas--to give medical personnel a way to view the body online as never before. In this interview, Brian D. Athey (right), the project's director, discusses the National Library of Medicine's creation of a digital image library; the future of anatomy and medical imaging; and the potential real-world applications the Visible Human Project presents for medicine and science. |
Fathom: What are the goals of the UM Visible Human Project (VHP)? |
 | |
| The goal of the UM VHP is the mapping of anatomic terminology and associated data onto the NLM Visible Human datasets. | |
Brian Athey: The general goals are to create an image database of anatomy that can be processed in various ways. Using labels of the structures and software to enhance the data, it will be possible in many cases to replace human cadaver dissection, and to create the next generation of anatomy training. The UM VHP will be utilized primarily by medical students, surgeons, dentists, nurses, paramedics, other health professionals and students in grades K through 12. |
Fathom: How did the NLM Visible Human Project get its start? |
Athey: The project came out of a 1986 long-range planning effort of the National Library of Medicine. They foresaw a coming era where the National Library of Medicine'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. One key aspect to the National Library of Medicine's goals is to create a standard image library. The idea was that if we had one dataset of digital human anatomy that everybody agreed to use, labs could then compare the performance of different algorithms and software packages on the standardized data. |
At the same time, in the late 1980s, it was known that anatomists at the Ph.D. level were not being trained the way they had been in the past. The molecular biology revolution was taking hold, and many people felt that anatomy as a research science had pretty much run its course. From an academic perspective, the field of human anatomy had lost its luster as an esteemed basic medical science. It was believed that since all the body parts had been identified, the questions of anatomy had been solved. This is a very narrow view of what anatomy is, and creates potential problems for the future. If you do not have people who are trained in anatomy, how are you going to teach it? |
Fathom: How did the Visible Human Project acquire digital images of the human body? |
Athey: Your chief qualification in the 19th century to become a professor of anatomy 
was to have a source of human tissue: a cadaver. The Visible Human Project obtained its source from one male and one female subject. In August 1993, convicted murderer Joseph Paul Jernigan (who had donated his body to science) was executed in Texas. MRI and CT data were collected from the cadaver, but first at a lower resolution, as that was all that we could capture technologically at the time. His body was frozen, cut into six blocks, and "sliced" in 1-millimeter increments at the University of Colorado. This process yielded over 1,800 cross-sectional digitized color images. Two years later a 59-year-old woman from Maryland died of heart failure, and her body, which had been donated to science, was "sliced" into 1/3-millimeter increments. These bodies and their images became the foundation of the Visible Human Project. |
Also, the trend at the time was to move away from film into digital techniques that used the third dimension, such as with three-dimensional CT and MRI scanners. Digitizing slices of the human body requires large files, several megabytes even into the gigabytes, especially by the standards of the early 1990s. The initial Visible Human male data from color, MRI and CT images is over 15 gigabytes. At the time, this was a huge amount of data, and very expensive to work with. My biological structure lab at the University of Michigan was one of the few in the world that could handle such large amounts of data. |
Fathom: Your work with the Visible Human Project not only allows for navigation through a rendered human body, but also allows users to navigate through immense data on the human body. How does visual modeling enhance our understanding of data? |
Athey: In the last 15 years, a new field called scientific or medical visualization has evolved. It allows you to transform numerical data into visual data, so that individual forms can be easily recognized by the human mind. The key here is in cognitive processes. The human mind has a strong ability to process visual information, much stronger, for example, than to process a table of numbers. You do not have to teach kids to process visual information; they just know how to do it naturally. However, everyone has to go to school to learn how to do arithmetic. The Visible Human data is so big and so extensive (15 gigabytes for the male and 50 gigabytes for the female) that many of us do not have the kind of memory that allows us to imagine how the body appears simultaneously in various planes of view. The opportunity to visualize anatomy in this way was the first contribution we and several others made to the field of the Visible Human. |
Fathom: What sort of visualization tools are available and in use? |
 | |
| A unique tool for anatomy instruction, the volume browser displays 3-D surface models, 2-D imagery and associated information. | |
Athey: We have been developing a tool--what we call a browser--that would allow you to cut the Visible Human data at various angles, giving interior views of the human body from any direction and perspective. Cross-sectional anatomy is good, but it does not allow you to examine the body from multiple perspectives, such as from a sagittal or transverse point of view. So we moved our thinking away from cross-sectional anatomy, to develop a browser tool that could give you views that essentially reconstruct the body. Recall that the data collection process for Visible Human was destructive; the bodies were destroyed by the procedure. What we have the browser accomplish is to heal this through reconstruction. This takes place in a computer, which can mold data together, and allows users to move through an integrated body. Let's say you want to navigate along some region of interest such as the aorta, a structure that branches and curves. You would be able to do so using a browser tool. |
One such tool, the Edgewarp browser, which existed before the University of Michigan Visible Human Project became involved, has the ability to make comparisons and morphological changes between known and unknown datasets. It allows you to basically warp and then fuse data. If you want to fuse a 3-D image of two people, the first thing you would have to do is some image stretching and compression. The way to do that is to identify points that we all share, like the tip of our nose or our earlobes: so-called landmark points. When you map these points, you will eventually be able to render a transformation. |
This capability could be valuable to us in several ways. For example, you could take datasets related to the human body that were not prepared as the Visible Human had been, but from other sources such as a high-resolution spiral CT image--a whole body scan, which you can have done in a doctor's office or a clinic. With the Edgewarp browser, you can actually take in this data from the outside and compare it to the Visible Human for reference and analysis. |
We used the Edgewarp browser and then added features to allow the tool to become usable over Internet networks, which makes it possible to interact with visible databases that could be anywhere in the world. |
Fathom: The development of browser tools would allow someone to reconstruct the human body over and over again. How might this be applied in medicine? |
Athey: One great example is that medical personnel could practice a surgical procedure utilizing the Visible Human data and an appropriate surgical simulation tool. This is very important for training surgeons on minimally invasive surgery, such as using a laparoscope. Laparoscopy is a newer technology that uses a small video camera and a few customized instruments to perform surgery with minimal tissue injury. This type of procedure requires a lot of practice in order for a surgeon to be able to perform it properly. So one of the great future uses of the Visible Human will be to allow for simulated surgeries. |
Fathom: Is this reminiscent of the surgical theater of old--a teaching arena? |
Athey: Yes, that is right. The anatomical theater of the 19th century was essentially an amphitheater where a professor pointed at the object that was dissected while students looked from above. The surgical theater today is very similar, except that you are actually doing something to a living human body. Thus the physicians who need anatomical training the most are the surgeons. Surgical anatomy, it is important to note, is very different than classical dissection because you have to be concerned about what you cut. If a surgeon wants to look at something in the dissection lab, he or she can just cut though the body and find that section, but everything up to that point is more or less destroyed. Surgery obviously requires much greater care, and so being able to simulate surgery with the Visible Human Project is a real benefit. |
 Another application here is for patient education. For example, a doctor might be able to show a patient where her gall bladder is and what a possible surgical procedure might entail. We imagine in the very near future that we will have rendered these various forms in the Visible Human dataset so that with a mouse click, you can visualize different relationships and patterns within the body. |
Fathom: The Visible Human Project has been in development for many years. What theoretical approaches (e.g. pedagogical, anatomical, biological, mathematical) have formed the project's direction? Has there been a marriage of multiple approaches toward a common goal? |
Athey: There has been an evolution. It is important to note that the Visible Human data was delivered to the National Library of Medicine for worldwide distribution. There are over 16,000 licenses to the Visible Human data out there, from countries all over the world, including Japan, the former Soviet Union, and South America. Very different groups utilize this dataset for various things. |
Teachers of anatomy started to use the data in a limited way. For example, laser disks were created to provide learners with cross-sectional anatomy. Many teachers simply processed the data in three dimensions and tried to use the cross-sectional anatomy that came from these pictures as a supplement for anatomy courses; for example rendering the esophagus or the spine. Then to improve on this, leaders in the field started doing what we would call flythroughs. These views enable you to visualize actually moving through various rendered objects from a changing point of view, for instance, navigation through the convoluted turns of the inner ear. |
We worked very closely with anatomy professors. But since this is a new digital medium, you need to do more than just work with the anatomy professors, as they are not always familiar with the technology. Technologists, like myself, became involved to make sure that pedagogical requirements were met. This is important not only from the point of view of what is taught, but also to ensure that information is delivered effectively for different kinds of students and different ways of learning. |
Fathom: How did mathematical and computational science inform the project? |
Athey: There are several approaches to applying math to the Visible Human data. The challenge is to try to understand some of the rendered objects in the VH data in geometric and mathematical terms. We have the privilege to be working with Dr. Fred Bookstein, who is an expert in a field called morphometrics and who has developed a theory of measurement to describe growth or change in biological structures, as well as the structures themselves, geometrically. This will be profoundly useful for the field of structural biology and human anatomy into the future, which very much needs new ways to describe our anatomy. It is not enough really, as was done in the days in the past, to just delineate the parts, if you will, and name them in Latin. We now know that the way the parts work together could be more important than the parts themselves. This is something for which mathematical modeling can be useful. Dr. Bookstein has really helped us become leaders in that area by applying his insights and discoveries in the emerging field of quantitative morphometrics to the Visible Human Project. This has been a great help to our understanding how the structures of the body fit together, and their natural and unnatural variations of structure and function. |
One limitation of the Visible Human data is that it is not alive; it does not move, and it does not have blood or any dynamic physiology to speak of. However, there is no reason that the Visible Human could not gain this dimension, especially as we have the power of the computer, morphometics and algorithms to model the physiology and to help enhance our experience. |
We are also enhancing the biological representation of the Visible Human by fusing different pieces of information with it, including imagery from different scales, such as three-dimensional images from microscopes. As an example, we have developed the ability with our laser scanning focal microscope system to make detailed three-dimensional images of the inner ear. There are possibilities, especially with Dr. Bookstein's software and algorithms, to properly fuse these datasets with the Visible Human data so that, when you locate the place in the ear where that data are supposed to be, you can actually go down and change scale and go from the millimeter to the micron, or jump down a factor of 1,000 and add a higher resolution look. This development would offer a more detailed look at the anatomy in any given place. |
Fathom: Do researchers who are utilizing the Visual Human datasets feed their results back into a central repository? For example, if researchers in Hong Kong were working on the human heart, would they be able to share their results with researchers here at the University of Michigan? |
Athey: This is the biggest and most important challenge right now. We are working very hard in developing our projects to make sure that scientists can get involved and share data. On the Internet, everybody puts his or her own little thing online. But how do you string these pieces together to make a story out of a bunch of Web objects? It is a big problem. |
What we want to do is develop tools that can allow for lots of different people to interact with the Visible Human data, add value to it, and put it into a common repository. A good example is anatomical labeling. What I am planning right now is to put a tool out on the Internet linked to a publicly available database that will allow anybody in the world to segment a piece of the Visible Human in three dimensions, cut it out, identify it, and name it in any language they want. For example, we are collaborating with a Chinese anatomist who is helping to finish our labeling both in Chinese as well as English. |
Fathom: What does the future hold for medical imaging technology? What other sorts of applications do you see coming out of the Visible Human Project? |
Athey: We envision a great number of applications. One important result of our work is that we are now allowed to visualize relationships and structures in a way that had been previously almost impossible to do, mainly because of the curved or oblique nature of the relationships that one wanted to see. The spinal cord is a prime example. If a researcher wanted to look at the whole spinal cord at one time, he or she would find that this is not easy to do because the spinal cord is curved with the backbone. However, if you could trace a line along the spine, when the anatomical data is digitized you can fit the image data to that line and essentially straighten it out and see the whole spinal cord in one picture. |
The future will see us working to obtain more Visible Human datasets and to improve on its resolution so that we can see important nerves and other small structures. But there is no question that the visualization technology that we are developing will allow us to see relationships between objects that we have never really been able to see before. This could lead to some very important fundamental changes in what kind of questions we ask in medicine. If you can see different relationships and symmetries clearly, it can provide you with new insights into the function and how it relates to genetics and pathology. |
|
| |
