When intensity modulated radiation therapy (IMRT) was first put into use at Houston’s Methodist Hospital in 1994, it forever changed how cancer patients around the world receive radiation therapy. Instead of a single radiation beam that treats the entire area around the tumor, IMRT uses a more precise multi-beam method to better pinpoint cancer cells in the body and spare surrounding tissue.

As advanced as IMRT is over previous methods of radiation therapy, however, the technology could not make up for the fact that each physician draws a different target for treatment based on their own anatomical understanding.

“In the old days, you drew a box around a target and you couldn’t deliver a dose higher than what normal tissue could bear,” recalls L. Anne Hayman, MD, neuroradiologist, Medical Clinic of Houston, adding that with the higher doses delivered at more carefully delineated targets, all surrounding structures must be evaluated.

Hayman also notes that radiation oncologists, who bear the primary responsibility for drawing the target, have not been trained in diagnostic imaging since the ’70s. “You are going out to the 7,000 radiotherapists and asking them to do IMRT, but how are you going to suddenly train them in diagnostic radiology?”

“We know that cancers spread in certain patterns along the lymphatics, but radiologists are not trained in that anatomy and we do not necessarily know where all the lymph nodes are on the CT scans,” adds Brian Butler, MD, radiation oncologist at Methodist. “Lymph nodes lie in spaces bound by structures, and these have been described by famous anatomists throughout history. But at a meeting of 17 different radiologists talking about where the targets should be for tonsil cancer, there will be 17 different opinions. We need consistency to draw a target for the radiation. If everyone draws differently, how can we show the benefit of IMRT across the board?”

Hayman recalls how treatment planning involved compiling information from a lot of disparate materials.

“I’d go to one clinical book that shows the anatomy from the front, then another that shows the body from the side, to map the most likely location of the lymphatics,” she says. “Pulling all the information together was a big part of my work.”

Hayman and Butler determined that if radiologists had a better understanding of the lymphatic system and radiation oncologists a better understanding of anatomy, treatment plans could be designed even more precisely. So they began pursuing a way to put that information at physicians’ fingertips. They first approached programmers at an Ivy League college, where they were told that it was impossible to create interactive technology that would merge years of established anatomic data with a 3D view of the patient’s body.

Undeterred, they turned to a group of 20-something Dallas-based video game programmers who jumped at the challenge. The result: proprietary technology that allows radiologists to readily call on a quarter century of hard-learned anatomical information to precisely analyze a tumor’s location and improve the ability of IMRT technology to target bad cells and spare healthy ones.


“It’s the incorporation of so much anatomic data that really makes this system so successful,” Butler says. “This is an enormous amount of data that we can apply directly to how we are treating our patients.”

“As a field, radiation oncology has no specific training in CT anatomy,” Hayman adds. “This helps us overcome that problem by having all the information about the human body already in the system.”

The system not only incorporates decades of anatomic discovery to help physicians map the human body, it also created a way to import the patient’s images. The sophisticated system works in tandem with the planning system via a monitor feed, so the radiation oncologist can merge real-time data with an historic understanding of, for instance, the lymphatic system. Once that data merges, the radiation oncologist can deliver a precise radiation treatment that considers the tumor size, location, growth pattern, and stage of the disease.

“Having access to all this data is great, but we knew that we needed to be able to integrate it with each patient’s actual anatomy,” Hayman says. “Radiologists are much more visual, they like to see things they can easily manipulate. Visual learning is so much more efficient than textual learning—it’s like reading a subway map versus actually seeing where the train goes. Seeing it makes it quicker to understand where we’re going and what we’re doing.”

The system now enable the radiologists to send 3D datasets to the workstation to identify where the tumor is in relation to where the normal structures are located. The designers used DICOM-RT in devising the system, which incorporates a workstation and individual servers on which resides software for each anatomic region. Head and neck and male pelvis are the first regions available.

“With the new technology, we can touch any structure on the axial CT with a mouse and the computer will tell us what that structure is,” Butler says. “Basically, every single pixel has a name associated with it, every structure is outlined and defined on CT. Then the computer comes up with the solution using an algorithm and tells us where it’s okay to deposit radiation and where it’s not okay.

“This is a revolutionary way to approach large amounts of data,” he continues. “Even if you don’t know the name of the anatomic detail, you can shoot at it with the mouse and text boxes of information appear to help guide your treatment.”

The program is highly interactive, and allows the user to assign anatomical structures a specific color: all network lymph nodes according to one classification can be assigned one color and arteries another, and colors can be switched at will as strategy shifts. It also employs the video game instant navigation strategy, with a body map at the top corner to show the user where they are in a stack of images.


Though they are careful to keep the identities of their young programmers secret, Butler and Hayman have no reservations when it comes to singing the praises of the gamers with whom they worked.

“These guys jumped in with both feet and were really open to our idea, which was to put images, text, graphics, and articles together and make it all electronic,” Hayman says. “To some, that concept might be daunting, but these younger guys didn’t know they couldn’t’ do it. As a result, these programmers did do it—and they are actually going to get some educational credit for their work at the University of Dallas too.”

Figure 1. In this screen capture, a 2D CT slice is test delineated with two different comparison lymphatic regions. The left-hand side of the 2D is faded out to drop down on top of a real-time patient CT. The right-hand side shows a normal anatomy side not faded out. All of these 2D slices can be adjusted to fit on to the patient CT. The 3D reconstruct shows a one-to-one relationship between the 2D slice and the volumetric 3D. (Click the image for a larger version.)

“They have really created out-of-the-future technology that incorporates color, panels that you can move or fade in and out,” Butler says. “They made good use of screen real estate and created a way to help physicians see clear, predelineated lymphatic targets and the anatomic details.

“As a result, physicians now have the confidence with this system to treat tumors with bigger doses while still sparing surrounding tissue,” Hayman adds. “When you have an atlas of anatomy accessible, that is a huge step in treatment potential. With our program, it’s like you can jump into a swimming pool and swim in any direction. It doesn’t limit you.”

The system’s expansiveness and ease of use are not just revolutionizing today’s patient care, they also have exciting implications for the future of treating all types of cancers. It also signals a potential expansion of IMRT into areas where physicians may not currently be comfortable due to their limited anatomic knowledge. The real next big step would be to teach the computer to overlay the anatomic atlas onto the patient automatically without the interface of human contact.

“The future benefits of this technology may be seen in cases of rectal and pancreatic carcinoma, lung cancer, prostate, and the lymphatics in the pelvis because we have essentially gone from a shotgun to a sniper rifle approach to killing malignant cells,” Butler says. “Imagine, too, if we knew where all the fibers are for speech and memory, and how that would impact treating cancers of the head and neck. This new system basically puts a sight on that rifle so we know exactly where to shoot.

“I have a 15-year-old, and I always thought video games were stupid, but strategy games really teach you how to think. Looked at this way, we can envision cancer as an enemy and radiation as a bomb dropped onto those enemies,” Butler adds. “This system takes targeted cancer therapy to another level because gaming technology pulls all the anatomic information together in one place and lets you do it easily.”

Hayman likewise sees the potential for integrating the system with other therapies as well, such as genetic profiling. That would help physicians know what types of therapy a cancer is likely to respond to and what types of therapy would best be avoided. As a result, physicians could customize a treatment plan that is based on the probability of what type of chemotherapy agent will be most beneficial to the patient, and they would be able to know what doses have the lowest probability of side effects.

“As exciting as this all is even today, from what the gamers tell me they have just started,” Hayman says. “They have all sorts of plans, and they say the technology will only get a whole lot cooler.”

Elizabeth Finch is a contributing writer for Decisions in Axis Imaging News.