A Stunning Look Inside the Human Body

By Grant Ellis

Richard Breiman’s fascination with medical imaging goes way back – to the time when computed tomography (CT) was in its infancy.

Following his graduation from medical school at the University of California San Francisco (UCSF), Breiman was training to be an orthopedic surgeon when the first CT head scanners were released by their inventor, the EMI Corporation in England. Breiman was so excited by their potential that he switched specialties and went to Stanford to train as a radiologist. While he was there he worked with Varian Associates on the development of one of the first CT body scanners – easily the fastest in the world at the time.

Today Breiman heads up the Henry I. Goldberg Center for Advanced Imaging Education at UCSF, creating spectacular 3D volumetric images as you see here. We asked Professor Breiman to tell us about the future of 3D imaging techniques.

How has 3D imaging evolved over the past 30 years?

In the 1980s, 3D image processing was very computer-intensive. We’d spend 45 minutes doing the computation for a single low-quality image from a stack of gray-scale CT scan slices that were ten millimeters thick. The scanners took one slice at a time. Today at UCSF we have CT scanners that take 64 images just 0.625mm thick in a single pass. They can take about 800 images in 10 seconds. But the images are still 2D, and they are still gray-scale. It’s not efficient for radiologists to view hundreds or even thousands of 2D slices one by one to assess results and understand complex anatomic relationships between abnormalities and normal structures.

So the vendors of scanning systems began offering workstations that convert stacks of CT gray-scale images into colored 3D images that are much more useful for diagnosis or surgical planning. Now an off-the-shelf Mac can generate high-resolution images directly from a CT scan. These images, as you see, do a superb job of displaying anatomy and abnormalities.

How do you create these remarkable images?

We start with CT images from one of the Radiology Department’s 64-slice CT scanners. The images, in DICOM data format, are routed from the scanner to a picture archiving and communication system (PACS) and can be sent directly to a 3D workstation for processing by a technologist. I use the workstation to burn a CD or a DVD containing the 2D images I want, and load them on my Mac.

We use innovative new DICOM High Definition Volume Rendering® software from Fovia to create 3D images from CT scan data. Leopard gives us 64-bit processing and allows us to handle a large number of images in an efficient way. I have used a Mac Pro and Fovia’s software to give demonstrations of real-time 3D volumetric processing using datasets of 2300 images. I could do the same thing with 4000 image slices if required.

Apart from their stunning visual effect, why are these images important?

Medical professional students and postgraduate trainees need to learn human anatomy thoroughly and efficiently prior to caring for patients, and they need to be able to review anatomy when needed long after their anatomy classes. People are 3D objects, but most of our anatomy atlases and diagnostic imaging examinations today are displayed in 2D. If you're trying to understand human anatomy by looking at 2D anatomic drawings or a stack of 2D cross-sectional images, you may not get it right. The ability to display medical images in 3D vastly improves our ability to teach anatomy, perceive abnormalities, understand their significance, plan for their treatment, simulate interventional procedures, and guide surgery.

About 20% of our routine CT scans get some sort of added 3D treatment. In some cases – aneurysms, for example – 3D imaging is automatic. In other cases, we may decide for diagnostic reasons or to better communicate the results to a clinician, to augment a scan with 3D renderings. This can be done at our diagnostic workstation or sent to a more specialized 3D workstation or to the operating room for use by a surgeon to plan or guide surgery.

We’re using 3D images in more applications all the time. Say the patient has an aortic aneurysm; we automatically do 3D. We take measurements and look at the vessels branching off from the aorta to help in planning the surgery. We use 3D imagery to assess the organs of transplant donors. We fly through the colon in virtual colonoscopies. Everybody who sees these images says, “But where’s the camera?” It’s hard to believe no camera is involved because the images are so lifelike, and that depends on what imaging tool you use. Clearly Fovia is an industry leader in the way they have approached this challenge of rendering large volumetric 3D images.

In one recent case, using High Definition Volume Rendering® software, I found a 6mm colon polyp that was missed when the 2D images were reviewed. There’s no way you could miss it in the 3D images. We adjust the lighting and the polyp just pops out.

We’re also using Fovia’s software and the Mac to create interactive 3D models for teaching anatomy. Students can click on lifelike images to see text boxes, highlight areas of interest, or rotate anatomy for a better look.

How has the technology you're using improved the use of 3D images?

In the past, we’d push the scan data to a remote workstation, and a technologist would push a 3D rendering back to us four hours later or even the next day. The best of all worlds would be if the radiologists could do it themselves on the fly. This technology lets them do that.

Another advantage is cost. Fovia’s 3D solution is software only. It doesn’t require the expensive workstations offered by imaging hardware vendors. It runs on the Mac and lets us manipulate superb high-resolution 3D images in real time. We have interfaced it with 3DConnexion SpaceNavigator, which gives us six degrees of freedom to fly through the images.

We’re using this technology more and more to plan surgery, or as an interoperative guide during the operation, or to teach, or to guide robotic surgical devices. I can’t do what I’m doing with any other technology.