Stanford Lab Emerges as 3D Imaging Leader

Laura Pierce, M.P.A., R.T.(CT) Stanford University

Sandy Napel, Ph.D. Stanford University

Geoffrey D. Rubin, M.D. Stanford University

During the course of producing extraordinary clinical images for patients and clinicians, the 3D Imaging Laboratory at Stanford University School of Medicine has established itself an international epicenter for developing and teaching the 3D image postprocessing techniques that are becoming increasingly critical to clinicians and researchers worldwide.

"On the research side, we're always working to develop new and more efficient postprocessing techniques," said Laura Pierce, M.P.A., R.T.(CT), manager of the laboratory. "On the educational side, we disseminate the information we have acquired here to the world so that other sites can use this technology and implement their own 3D laboratories."

The lab recently unveiled its very latest in 3D images to the public at Flickr.com/photos/StanfordMedicine.

Pierce has been involved with the lab since its inception in 1996, along with Stanford radiology professors Sandy Napel, Ph.D., and Geoffrey D. Rubin, M.D. The lab was among the first to develop advanced visualization for CT colonoscopy as well as vascular visualization techniques including removal of bone (segmentation); maximum intensity projections; curved planar reformats that follow the trajectory of vessels through the body for display on a single image; and quantification, such as measuring the maximum diameter of an aneurysm or the position of a stent graft and measuring change over time, according to Pierce. Listen in as Dr. Rubin details the lab's background and postprocessing procedures.

The lab has also developed several computer-aided detection (CAD) techniques for procedures such as CT colonoscopy and lung nodule detection. Data used to create 3D images comes from images produced by CT and MR imaging scanners. Examinations appropriate for 3D imaging are routed to the lab via PACS. The majority of image processing is done by specially trained radiologic technologists.

"In our past lives we were technologists in CT, MR and cath-angio," said Pierce. "There are seven of us processing cases full time and our volume has increased to almost 1,000 exams a month."

Protocols Drive Postprocessing

Images are processed according to protocols based on the type of imaging study, Dr. Rubin said. "Some focus predominantly on measurements made in the dataset, others focus on visualization and the creation of images and many have elements of both," he said.

The lab currently has about 90 protocols and more are continually being added, said Dr. Rubin. These protocols enable the quantitative measurement that is essential for quality assurance, he said.

"Although every exam is slotted into a specific protocol we have enough breadth in our protocols to accommodate the wide range of clinical scenarios we may encounter," said Dr. Rubin. "The protocols form the basis for a highly formalized quality assurance/improvement program. We have learned lessons that go well beyond the 3D lab and could benefit many areas of diagnosis, both within radiology and beyond." Listen in as Dr. Rubin explains opportunities for quality improvement in a 3D imaging program.

After creating protocol images, the technologist also produces images unique to each patient's dataset, said Pierce.
"The technologist has to look into the patient's history and understand what the radiologist needs to see," she said. "For example, if the patient has a neoplasm in the pancreas, the technologist will decide which views will best demonstrate that neoplasm to allow the radiologist and the referring physician to fully characterize its location and extent."

3D Postprocessing Evolves with CT

The mathematical basis for 3D postprocessing has been around since the early days of the first CT scanners, with significant gains coming recently, Dr. Napel said.

"Over the past five years, the clarity and detail in these images have improved substantially and most of that is due to the improvement in resolution in the imaging devices," he said. "When CT went from single to multidetector, and it became possible to get very thin slices, the ability to create these images was improved."

Due to advancements in CT and available datasets, 3D imaging has become a necessary part of daily practice, according to Elliot K. Fishman, M.D. "It's no longer an option," said Dr. Fishman, a professor of radiology and oncology at The Johns Hopkins University School of Medicine and director of Diagnostic Imaging and Body CT at The Johns Hopkins Hospital. "We used to have 100 slices and now we have 1,000. We can't begin to look at all of those slices without 3D."

Equipment vendors have been working for the past several years to improve data rendering techniques, said Dr. Napel, resulting in fine enhancements such as imparting "light" to the images. "An image can be shaded in such a way as if there were, for instance, a light sitting off to one side, so you can better see surface variations the way you would with room lighting," he said.

Postprocessing speed has continued to improve as well. "For perspective, the first virtual colonoscopy images that we made in 1994 were rendered on a $250,000 computer and it took 48 hours to create 1,000 frames of a movie to fly through," Dr. Napel said. "Now you can do that in real time as you drive through the volume with a computer that costs no more than an average laptop. Progress has been facilitated through faster processors and clever software implementations."

Because some added time and cost are still involved, only about a tenth of the CT and MR examinations performed at Stanford are processed as 3D images, Pierce noted. "We only use 3D when it's going to add value to a patient's exam," she explained. "I think we have to be good stewards of this technology and not increase patient cost needlessly."

Flickr Site Aids Patient Understanding

The 3D photos on Flickr include a sampling of images ranging from vascular to musculoskeletal. A few viewers have posted comments and questions—answered by Stanford staff—about topics like avoiding arterial calcification and radiation exposure from CT scans.

"The site is mainly for the public to see and appreciate what we do," said Dr. Rubin. "The opportunity for public education is compelling."

The ability of 3D images to demonstrate anatomical features plainly to even the layperson is beneficial to practitioners and patients alike, added Dr. Napel. Referring physicians can use the images as visual aids to help patients understand the disease process. "It's a lot easier for patients to understand what's really going on inside their body when they see these images," he said.

Above, from left: A CT image from Stanford's 3D Radiology Laboratory illustrates cranial sutures holding a piece of bone removed during surgery; a lateral view of the upper torso, constructed from CT data in Stanford's 3D Radiology Laboratory, reveals the lungs in relation to surrounding; and a CT image of legs reveals underlying bone, muscle and vasculature in this 360° composition from the laboratory.Images courtesy of Stanford Radiology 3D Laboratory.

3D in Every Institution is Goal

Although understanding the nuances of visualization requires highly specialized skills, few such training opportunities for technologists exist, according to Pierce.

"Right now the only training a technologist can get is from the vendors, and it's only for a few days once you purchase a workstation," she said. "The vendors don't really have any idea about clinical images that are necessary to display pathology." Few academic programs offer 3D training, she added.

Stanford's 3D lab offers clinical training on postprocessing skills via fellowship and assistance to institutions interested in starting a 3D lab, said Pierce. While Stanford may have emerged as a trendsetter, Pierce emphasized it's possible for any institution to create such a lab.

"What we do here is very reproducible," she said. "It's not that we have the great technology here. You can duplicate it at any site, anywhere around the world.

Dr. Fishman agrees. "Although Stanford is a role model of excellence in terms of 3D imaging, everyone should be doing it," he said. "Then it's just a question of the method. There are different models at different places, but when all is said and done, the important thing is that you use 3D imaging."

"I would like there to be 3D labs in every radiology department, even if it's very small," said Pierce.