Sponsored by an Educational Grant from Varian Medical Systems
by LISA FRATT
Over the last 5 years, advances in image acquisition, storage, and transfer have dramatically impacted how radiation oncology departments work. No longer does treatment hinge on a few images of questionable quality but on a greater variety of high-quality images.
Several years ago, most radiation oncology departments depended on the generosity of diagnostic radiology. Without a dedicated CT scanner, radiation oncology staff had to borrow time on a diagnostic CT scanner. Arthur Boyer, PhD, professor at Stanford University School of Medicine (Stanford, Calif) and director of the radiation physics division of the radiation oncology department, recalls, “Five to 15 years ago, radiation therapy was sort of a boutique specialty. Even large departments treated each patient as a special case. There wasn’t a great deal of efficiency.”
Having images available throughout the clinical process makes a tremendous difference in how patients are being treated today. Whether in treatment planning or treatment delivery, clinicians rely heavily on images to make decisions regarding radiation therapy treatment protocols.
Fast-forward to 2003. A snapshot of today’s radiation oncology shows a distinctly different department. For example, most departments now have their own CT scanner and handle images with CT simulation. Consequently, radiation oncologists and medical physicists are inundated with scores of high-quality images for each patient. Images aren’t limited solely to CT; most departments handle a fair share of MRI and PET images as well. At the same time, patient volume is exploding. Boyer says, “The sea change has occurred. It’s the age of baby boomers, and radiation therapy has increased in treatment volume so dramatically that departments have to become efficient production-line operations.”
The future will bring additional changes to radiation oncology departments. Walter Grant, PhD, associate professor of medicine at Baylor College of Medicine (Houston), predicts, “MR, PET, and CT will be more frequently used, and PET-CT is the hot rage. I believe we’ll see PET-CT scanners in radiation therapy departments in 5 years.” This could allow departments to better tailor treatment, but it also requires medical imaging expertise not readily available in most radiation oncology departments. Medical physicists and radiation oncologists are keeping their fingers crossed for some time-saving breakthroughs as well. Autosegmentation, which draws anatomical structures on the images with the push of a button, is high on the wish list.
Breakthroughs in image management are coming at seemingly breakneck speed; however, one advance seems to serve as the glue that holds the department together. Grant says, “We would be lost without DICOM RT. It’s part of the reason we’ve seen acceleration in radiotherapy in the last 10 years. The reason is two-fold. DICOM RT is the solution to a lot of problems. We’re dealing with lots of competing vendors and lots of subspecialties. The ability to push images without special software is truly a blessing. If we ever implement DICOM RT to its fullest level, it will take away the fear of implementing new technology.”
Images in PlanningImaging in radiation therapy treatment planning has improved substantially in the last few years. Radiation therapy departments have moved from using film, and sometimes suboptimal film not designed for the higher x-ray energies used in radiation oncology, and onto filmless operations. This shift has resulted in a number of benefits. Richard Emery, MS, chief physicist and director of radiation services for St. Vincent’s Comprehensive Cancer Center (New York City), recalls how the department functioned just a few short years ago. The process was entirely manual. Patients had different file folders and jackets, which were used to transport film around the department. Missing film and the large amount of real estate required for storage were major issues. Today, however, the department is filmless. Images, structure sets, dose information, and treatment history are stored on the server and available to any user within the department. The result? No more missing films. The department relies on Varian Medical System’s (Palo Alto, Calif) VARIS Vision system to manage images throughout the clinical process. Emery explains, “It’s one integrated system with minimal import and export transactions.” This, of course, streamlines workflow, but there are other benefits to digital, particularly in the patient care arena. Emery says, “We verify treatment by comparing an electronic reference image with images acquired at the time of treatment. These images can be reviewed in real-time while the patient is on the table. This essentially amounts to on-the-spot, real-time review and analysis that allows us to deliver treatment accurately and confidently in minimal time while the patient is on the table.” A final advantage of digital imaging in radiation oncology is improved quality, says Emery. Images can be optimized and enhanced as necessary. Digital, however, is not the end of the line for improvements in treatment planning imaging. “We will use kV imaging to supplement existing electronic portal images. These have distinctly different [improved] image quality and may be complementary,” explains Emery. Moreover, kV imaging may allow real-time or continuous patient monitoring during treatment. This includes monitoring organ motion. These advances could bring radiation oncology closer to its lofty goal of making patient treatment more accurate by increasing setup reproducibility and decreasing respiratory-induced motion and positional uncertainty. Tumor localization will be improved, enabling radiation oncologists to treat smaller volumes and less normal anatomy. |
A Visit to the Past
Radiation oncology imaging has come far in the last decade. Ten years ago some departments routinely projected CT planning data on the back of a door. Radiation therapists magnified the images and drew organs by hand. Chuck Pelizzari, PhD, associate professor and director of the medical physics section in the department of radiation oncology at the University of Chicago (Chicago), recalls, “We started using CT images in the mid-1980s. There was no DICOM, so we had a different solution for every type of scanner. We had a homemade treatment planning system and homemade contouring program. It was really pretty painful.”
In this pre-DICOM era, images from a few diagnostic radiology scans would be transferred into the treatment planning system via a large floppy disk. The next step, says Grant, was transfer via DICOM. Naturally, there was a hitch. Older scanners could not produce DICOM output. A $10,000 black box connected to older CT scanners translated the output into DICOM for the treatment planning system.
PET/CT images are tools that help define gross tumor volumes.
Typically, radiation therapy departments used cross-sectional CT or MRI images for treatment planning. At most hospitals, this entailed borrowing hard-copy film images from diagnostic radiology. These were transferred manually to a simulation radiograph or radiation therapy planning CT. Jeff Michalski, MD, clinical director of radiation oncology and associate professor of radiation oncology at Washington University of St. Louis (St. Louis, Mo), admits, “Hard-copy transfer or digital transfer was prone to a lot of problems. The patient could be in completely different positions on the two images, or you might be using a preoperative scan and a postoperative scan. The patient could have had chemotherapy.” Each scenario creates a less than ideal image transfer situation.
The other not-so-minor issue under this scenario was access to a CT scanner. Few radiation oncology departments had their own CT scanner, so they borrowed time from diagnostic radiology. Grant notes, “This really put a limit in the number of patients to whom we could apply 3D treatment planning. We were usually allotted one slot per day from diagnostic radiology.” Radiation therapists needed to know exactly what they were doing in the treatment room prior to the CT scan because they had one shot at the scanner and this information had to be communicated to diagnostic radiology before the CT scan. Grant concludes, “It took a lot of effort and coordination to do one or two patients a day.”
Treatment verification was equally complex. The University of Chicago radiation oncology department started using portal imaging for treatment verification about 5 years ago. Pelizzari opines, “At that point, weekly port films were kind of painful because the software wasn’t well developed.” Still, most patients had a simulation film done; the simulation film was manually compared with a port film on a light box. This comparison could be a bit subjective, says Pelizzari.
“It’s not direct verification because you can only see things relative to bony anatomy,” Michalski adds.
During the late 1990s, most radiation oncology departments were dealing with relatively few images. A patient might have three to five CT slices for scanning and a few simulation films. Nevertheless, storing and handling images could be quite cumbersome. Boyer says, “Five years ago, film was the primary medium used to capture and acquire information. It was stored in a film package, stuck in vertical slots and shelves, and inevitably lost on someone’s desk, which resulted in the perennial film hunt around the department. It’s a ubiquitous story.” Images also were stored on magnetic tape.
All of these factors-few images stored on floppy disks; older, non-DICOM machines; relatively low quality images-inhibited growth in radiation oncology, says Grant.
Imaging Today
In the last few years, radiation oncology imaging has witnessed some fairly dramatic changes. Treatment plans are typically based on larger numbers of CT images, and other modalities, including MRI, PET, SPECT, and occasionally ultrasound, are employed on a much more frequent basis. Most radiation oncology departments have implemented CT simulation, so they no longer need to borrow time on diagnostic radiology CT scanners. These changes have affected the relationship between radiation oncology and diagnostic radiology.
Brad Murray, MSc, senior medical physicist at Cross Cancer (Edmonton, Alberta, Canada) explains, “Our relationship with radiology has changed to a certain extent now that we have our own CT simulator. Now we’re asking for access to their MRI and PET equipment.” Grant opines, “There’s actually a closer relationship between the two disciplines. It’s more of a doctor-doctor relationship now.”
Most radiation oncology departments have limited experience with the sophisticated images used in treatment planning, so staff may confer with diagnostic radiologists on treatment planning issues. For example, a neuroradiologist may be asked to help interpret a PET scan, or a radiologist may provide input on the exact location of a tumor.
The implementation of DICOM and CT simulation has eliminated many of the inefficiencies that plagued radiation oncology departments in the 1990s. Pelizzari says, “The process is completely transparent now. As soon as a CT scan is acquired, it is automatically sent to the workstation. Most contouring is done on the CT simulation workstation. The images and contours are sent to the treatment planning system via DICOM RT and are used in much more sophisticated and sensible ways.” For example, 3D contours can be used in the treatment plan, and density information from the CT scan can be used for dose calculation.
PET/CT images (center) are tools that aid in the simulation (right) in treatment planning (left).
Merging information from MRI scans into the treatment plans is fairly routine. Image fusion tools on the CT simulation workstation integrate the data. MRI is especially helpful in brain, head, neck, and central nervous system tumors. Michalski notes, “MRI can be a better diagnostic tool to define the target. We can transfer the images with the patient in the same position as for therapy and not lose the electron density information for dose calculation.”
Not all modalities can be integrated into treatment plans as smoothly as CT and MRI. Although PET can help radiation oncologists better identify diseased areas, fusing PET and CT can be a bit problematic on stand-alone units. Michalski says, “Transfer and fusion of PET images into the CT treatment plan can be a cumbersome process because the files are not in a standard DICOM format.”
Some departments have addressed this issue by writing their own translation software. Image fusion with PET-CT scans tends to be simpler because the CT component of the PET-CT scan is fused with the treatment planning CT. PET-CT also allows radiation oncologists to deliver more accurate treatment due to the fact that they now have morphologic and biologic information in a single image data set. Physicians can see areas that have disease and increase the dose to these areas; they also can exclude areas that do not show any FDG uptake.
CT, however, is not the only imaging modality that enables radiation oncologists to deliver more targeted treatment. Most radiation oncology departments rely on an increased number of MRI and PET images and thinner CT slices for treatment planning and delivery. Murray says, “During the entire process, we typically have 100 or more CT images, a similar number of MRI images, and portal images twice a day, so 250 to 350 images per patient are not uncommon. It is absolutely important to have thinner slices for target volume definition and critical structure definition.” A brain scan, for example, is completed in 1-mm slices.
Thinner slices, however, can be a bit of a double-edged sword. Pelizzari points out, “Thinner slices are nice for visualizing small structures. On the other hand, they are noisier and take more time.” And on the treatment planning end, thinner slices do correlate with more time spent identifying anatomical structures during the contouring process.
Radiation therapy is a two-part process-planning and treatment- that requires verification. Grant notes, “Image-guided planning requires image-guided verification.” Many departments have implemented electronic portal imaging on their treatment units. Portal images can be compared to a simulation film or digitally reconstructed radiograph. The primary advantage of an all-digital format is that the images can be compared in a quantitative way, says Pelizzari. The subjective uncertainty inherent in comparing two films on a lightbox is minimized. Michalski adds, “Electronic portal images can be taken more frequently. We can start to better understand the types of errors that occur in treatment delivery.”
Another type of treatment verification is real-time ultrasound, which is used in prostate cases. Radiation oncologists use this technique to localize the target and identify the bladder, prostate, and rectum before treatment is delivered and for more immediate verification. Michalski concludes, “Centers using real-time ultrasound report improved localization. The real advantage of improved localization is that we can start to reduce the margins added to treat for errors and radiate smaller amounts of normal tissue.”
Many radiation oncology departments are either weaning themselves off film or eliminating it altogether. Images are typically stored on a hard drive, tape, or optical disk. At Cross Cancer, everything is networked and stored electronically to the VARiS Vision database. VARiS Vision has resolved a number of issues. Lost images are no longer a problem, and images are available to a variety of people at any time. “One thing that is tremendously better is our ability to use all of this information in an electronic chart review process,” says Pelizzari of the University of Chicago. “We can easily pull up digitally reconstructed radiographs, treatment plans, and portal images.”
DICOM RT, CT simulation, PET, and other advances in imaging have truly benefited radiation oncology departments. DICOM RT is a perfect example of how technology has changed the way radiation oncology departments work. Pelizzari explains, “DICOM RT has made all the difference in the world. We’ve been able to stop thinking about tape format and focus on the information in images.”
Heading Into the Future in the Treatment RoomImaging in the radiation therapy treatment room parallels imaging in radiation oncology. Arthur Boyer, PhD, medical physicist and professor at Stanford University School of Medicine (Stanford, Calif), explains, “We’re just beginning to see the leading edge of electronic imaging and efficiencies of electronic imaging.” For example, the department does use PET in the treatment planning process, but software for handling images is still crude. This situation is universal across modalities. Boyer says, “The integration of all these systems into one large departmental system is a major challenge. Right now, the image handling process is not adequate to the volume of work that needs to be done.” Despite the gap between the present and the ideal, the department is well on its way to becoming a streamlined, efficient operation. DICOM and DICOM RT have been a big help, says Boyer. And the VARiS Vision Information & Image Management System, manufactured by Varian Medical Systems, Inc (Palo Alto, Calif), facilitates workflow. Boyer notes, “You need a PAC system designed for a radiation therapy department. A diagnostic PACS doesn’t work.” Why? The therapy PACS must provide a relatively small number of images and data of diagnostic type acquisitions and a wide variety of types of images such as portal images and dose distributions. Over the course of treatment, the radiation oncologist must access this information many times to manage treatment. The goal of diagnostic PACS, on the other hand, is to present a large amount of imaging studies acquired for a large number of patients to the radiologist a few times for diagnostic purposes. The next round of imaging breakthroughs will have a substantial effect on the treatment room. Boyer explains, “Portal and kV imaging in the treatment room will be the next big push. kV cone beam CT on the treatment machine is a revolutionary development; it will allow us to quantify the location of soft tissue. CT scanners in the treatment room are just a step toward cone beam CT on the treatment machine.” Other advances include kV fluoroscopy in the treatment venue, which may be used primarily for respiratory gating. Optical imaging, viewing the external body of the patient, may also facilitate respiratory gating in the future. Boyer concludes, “For these things to happen, we need enabling technologies. The most valuable quantity in radiation therapy is the radiation oncologist’s time. The radiation oncologist needs time to think and make judgments, not wiggle mouses. The department needs an electronic means of conserving that very precious commodity.” That’s a tall order and includes a range of enabling technologies, such as radiation therapy PACS, cone-beam reconstruction, and autosegmentation. There are also complementary shifts in human resources that can facilitate efficiency. The relationship between radiation oncology and diagnostic radiology needs to become even tighter because diagnostic radiologists’ experience analyzing 3D data sets has to be transferred to the radiation oncology practice. The results, improved localization, and consequent ability to treat less-normal tissue make the transition well worth the effort. |
Fast-Forward to the Future
Despite the flurry of advances, radiation oncology imaging technology is still evolving. “We will see increasing use of more and more different kinds of imaging,” Pelizzari predicts. This includes PET, PET-CT, multislice CT, and MRI. “Hopefully, this will generate more precise information about the definition of structures and targets,” he continues.
Two of the most promising modalities for target volume definition are PET and PET-CT. Grant says PET-CT is the hot rage and predicts that PET-CT scanners will be located in radiation therapy departments in the next 5 years. Like other advances, this requires adjustments on the part of radiation oncologists and physicists. Pelizzari explains, “It becomes more important that the patient is always in the right place. We will be shrinking margins and moving fields. It’s possible to make a disastrous mistake without accurate patient setup.” How can radiation oncologists avoid these errors? One way is to utilize their counterparts in diagnostic radiology. As more sophisticated images are implemented in radiation oncology, radiation oncologists and physicists will come to depend on the expertise of specialists to help them read and interpret images, says Grant.
PET and PET-CT scanners are just one part of the PET equation; the other factor is radiopharmaceuticals. Michalski says, “We’re going to see more PET agents to help identify and predict the behavior of tumors. These could allow us to adjust the volume or combine radiation therapy with agents that might selectively target hypoxic cells.”
PET isn’t the only imaging modality expected to impact radiation oncology in the future. The use of MRI scans also is expected to increase. According to Murray, “MRI is becoming almost like CT simulation with the added benefit of functional information. MRI simulation will become available and will be in the radiation therapy department. Due to the fact that MRI can provide both structural and functional information, planning systems will be able to plan with MRI scans to avoid fusion issues.”
With VARiS Vision, clinicians can control the full treatment delivery process and have images immediately available for review.
While imaging advances will certainly facilitate treatment planning, they also will impact treatment verification. Michalski says radiation therapy treatment verification is going to take off in a big way. In the next 5 years, there will be a move to real-time treatment verification such as B-mode ultrasound. Michalski continues, “There will be tools in the treatment room independent of or attached to the accelerator that will allow radiation therapists to see the tumor second by second and track and radiate it dynamically.”
These imaging advances have many implications for radiation oncology departments. One of the issues that departments will certainly face is an explosion in the amount of data they deal with. Let’s face it-one of the things about port films is that they are weekly and require a mere side-by-side comparison with a treatment planning CT. Michalski says, “In the future, we will be dealing with megabytes of data daily, and some of them will be real time.” How can radiation oncology effectively manage and utilize the data? Michalski believes the answer is computers. He says, “We’re going to use computers to help decide if treatment is going according to plan.” For example, if computer analysis demonstrates that treatment covers 98% of the target volume, and that meets the department’s threshold, then the plan will be implemented. If, on the other hand, computer analysis shows a figure below the department’s threshold, the plan must be modified. The upshot, says Michalski, is that radiation oncology will need the assistance of IT.
Radiation oncology departments have developed a healthy wish list to take them into the future. Grant says, “Autosegmentation-the ability to push a button and have everything drawn automatically on 150 images-is a big thing.” Murray agrees and adds it to the list. “Right now, we have multiple data sets stored in multiple places. Everything is stored electronically, but sometimes it’s stored in too many places. We would be better off with a central data repository that all systems could access.”
What will a cutting-edge radiation oncology department look like in the next 5 years? It will very likely use its own PET or PET-CT scanner, and it may have an MRI simulator. The department will increasingly rely on the expertise of diagnostic radiologists for image interpretation. Radiation therapists will be able to more accurately target, deliver, and verify treatment, and computer software may facilitate these tasks and assist with image management. And for sure, it will be a streamlined and efficient operation. This is absolutely necessary, says Boyer. That’s because patient volume is expected to increase over the next 20 years.
