Radiologists use 3D tools to work smarter, not harder

Radiologists' transition from reliance on axial views to 3D interpretation of multislice CT data is not a question of if, but when. The increasing data load of 16- and 64-slice scanners challenges efficiency, as does the rise in imaging volume overall. The need to interpret more cases in less time will continue to put pressure on radiologists. But users at a range of sites, from academic centers to private practice, say 3D techniques are one way to work smarter instead of harder.

Many radiologists have yet to make the leap to 3D. The reasons relate in part to technology, perception, and training. Three-D workstations can be difficult to learn and use, and workflow integration is often poor. Training does not always focus on the most useful aspects of 3D interpretation. And some radiologists perceive 3D applications as too specialized for routine practice.

That attitude may change as more users recognize the value of 3D in their everyday work, be it CT of the chest or musculoskeletal and emergency room applications. Users who have adopted systems ranging from PACS-compatible 3D software to stand-alone workstations say 3D adds critical clinical information.

"Using 3D software with our PACS workstation has completely modified our workflow," said Professor Alain Blum, chief of radiology at Centre Hospitalier Universitaire in Nancy, France. "It has been adopted by our entire team, from senior radiologists to residents and technicians, as postprocessing becomes available everywhere in our department."

At Brigham and Women's Hospital in Boston, use of multiplanar reformats for review of CT data is integral to everyday practice. MPRs are increasingly considered the standard of care for CT data interpretation, according to Dr. Matthew Barish, director of the 3D and image processing lab at Brigham. Multislice CT users who are still relying on 5-mm-thick axial sections and not reviewing sagittal or coronal reconstructions are wasting the power of their high-resolution scanners.

"Radiologists need to realize that the standard of care has changed and that more diagnoses are more apparent from 3D images," he said.

At Brigham, up to 80% of all CT images are interpreted using MPRs, with the remainder split between maximum intensity projection (MIP) images and volume-rendered data sets for angiography. These two complementary techniques are typically used together.

Not only are MPRs quick to create on the fly, the capability to do so resides on many existing PACS workstations, said Dr. Shawn Teague, an assistant professor of radiology at Indiana University School of Medicine in Indianapolis.

"MPRs add very little to your reading time, and sometimes they can actually decrease your reading time," he said. "If it is a problem-solving situation, you can figure something out much more quickly from an MPR than if you had to struggle through axial images multiple times."

ROUTINE VIEWS

Three-D views offer numerous clinical advantages to radiologists in their everyday practice, with applications that begin with but are not limited to vascular, neurological, musculoskeletal, abdominal, and trauma cases. From simple 3D visualization to dedicated applications in cardiac imaging and the colon, radiologists find the technology invaluable for describing complex anatomy.

In musculoskeletal imaging, everyone benefits from 3D-surgeons, radiologists, and patients, said Dr. Georges El-Khoury, director of musculoskeletal radiology at the University of Iowa Hospital and Clinics. Three-D is particularly useful in understanding complex injuries in the acetabulum, knee, ankle/foot, and spine; assessing the integrity of fixation devices; and evaluating tendon abnormalities and soft-tissue tumors.

Barish also finds 3D helpful in musculoskeletal applications. Classifying complex fractures or comprehending the complexity of individual fragments associated with the fracture pattern is easier with 3D, he said.

For Dr. Mark Herbst, a solo radiologist who provides interpretation services for 40 imaging centers, 3D is an accurate way to localize lesions in multiple planes. It's a feature he finds helpful on a day-to-day basis.

"If you see a lesion that is questionable, all you need to do is click on it, and, depending on the program you are using, it will synchronize and show all the planes with the same dot," he said. "You can then determine whether it looks real to you or not, or if it is artifactual or volume averaging, for example. I find this really boosts my confidence in my findings."

TRAUMA AND ER

Volume-rendered techniques are also useful in ER and trauma applications, providing the ability to quickly create images that convey important information in just a few minutes. Three-D interpretation goes to bat in both trauma and nontrauma cases, shedding diagnostic light on everything from displaced fractures to aortic trauma, diaphragm injuries and rupture, cerebrovascular trauma, and stroke.

But 3D is only as valuable as it is available, said Dr. Robert Novelline, director of emergency radiology at Massachusetts General Hospital, who spoke at the Symposium on Multidetector-Row CT in San Francisco this summer. He advocates radiologist availability to interrogate and interpret 3D views on a 24/7 basis.

Specific 3D views most useful in ER applications include:

- surface rendering- representing the majority of 3D in the ER, it is excellent for fractures and fracture-dislocations, vascular trauma, and vascular emergencies;

- MIPs-help visualize blood vessels adjacent to bony structures; and

- volume rendering-more informative than surface models, volume rendering reveals the interior structure of organs.

"Three-D views can determine who is a surgical candidate," Novelline said. "More important, they can also determine who is a challenging surgical candidate and who should receive a stent, and they provide an incredible amount of diagnostic information."

CT COLONOSCOPY

In addition to generating volumetric views for general applications, 3D workstations are increasingly equipped with dedicated applications driven by clinical problems for use with CT angiography, cardiac CT, and CT colonoscopy, among others. These packages rely on powerful automated tools for bone removal, segmentation, vessel tracking, aneurysm analysis, and stent-graft analysis.

In the colon, for example, vendors continue to refine 3D views that allow multimodal demonstrations of data simultaneously, coregistered for optimal lesion assessment. Newly introduced 3D software supports not just visualization but also detection of pathologies in the colon, which could bolster use of CT colonography for cancer screening.

Typical 3D features used for CT colonography include automated centerline and organ dissection, high-quality endoscopy, and matching prone and supine data. Among future applications are electronic cleansing and tagged stool removal tools, as well as automated polyp marking and analysis features.

Many virtual colonoscopy experts consider primary axial views with 3D for problem solving to be the optimal method for reading exams, according to an informal survey conducted at the International Symposium on Virtual Colonoscopy last year. But with several studies indicating higher sensitivity with 3D primary reads, not all experts are ready to endorse the 2D primary-read model.

Barish, who organizes the annual conference and has also developed an onsite training program for CT colonoscopy, noted that the best approach for now is a combination of 2D and 3D.

VASCULAR APPLICATIONS

In the vascular system, dedicated 3D features include automated organ dissection, calcium analysis packages, coronary vessel tracking, and detailed vessel depiction. Such tools are designed to improve productivity by increasing the speed with which radiologists can identify and report key findings on CTA and MRA studies.

"The vessel analysis package is very efficient for CT vascular studies such as aortic evaluation or runoff," Blum said. "With the module that we are using, a complete evaluation of the vascular tree is available with the relevant data and report within 30 minutes after an exam."

An additional advantage of 3D in CTA is visualization of small renal arteries or accessory renal veins that are often overlooked, Barish said.

"I do abdominal imaging, and vascular anomalies of the mesenteric vessels that become easily apparent are easy to overlook in the absence of 3D processing," he said.

At Indiana University, Teague relies on volume rendering to evaluate the spectrum of cardiac and vascular cases.

"MPRs are going to be the mainstay as far as making a diagnosis, but in an aorta case, for example, we always use a volume-rendered image," he said. "We are able to put points on that image in an MPR setting, creating a centerline down the vessel. No matter how the vessel is curving in the body, we have a centerline, and images are automatically generated that are perpendicular to that line. As a result, we have true cross-sectional images of the aorta."

Such a tool is useful because it allows radiologists to more accurately measure the aorta, assess whether dilation exists, and identify aneurysms.

Volume rendering is also used for pulmonary vein mapping, providing an overview of the anatomy for cardiologists planning ablation, Teague said. In cases involving cardiac tumor, for example, cardiothoracic surgeons appreciate volume-rendered images because of the way anatomy is presented in overview.

"When they go in for surgery, they have a very good idea of what to expect," he said. "For cardiovascular, and even for vascular applications outside of the chest, you are going to make some kind of 3D reformat for every single one of those cases," he said.

THE BRAIN IN 3D

In the brain, 3D tools pinpoint aneurysms and provide detailed mapping information for surgical planning. For example, 3D color volume-rendered images can confirm the presence of bilateral internal carotid artery aneurysms, as well as appropriate management of those aneurysms by neurosurgical or endovascular techniques.

Herbst uses 3D in MR and CT brain imaging applications, finding it helpful for identifying signs of early traumatic brain injury in auto crash victims.

"The new MR machines are very sensitive to areas of hemorrhage, and with 3D software I can measure the brain plus cerebrospinal fluid volume, subtract the CSF volume, and come up with the brain volume measurement," he said. "It is amazing how easy and reproducible this is. I can do it in five or 10 minutes."

Such volume measurements-now also available on some 3D workstations for lung volume measurement as well as brain CSF-are valuable whether acquired using CT or MR. Volume data can be extracted easily from today's MSCT images, according to Herbst.

"With the software available now, you click on the nodule, then click on the 'measure volume' button, and you are done," he said.

Users do need to be aware that such volume measurements are subject to plus and minus error. At a 3D workstation face-off event held as part of the Symposium on Multidetector-Row CT earlier this year, automated volumetric analysis tools for measuring both lung nodule volumes and cerebral blood flow in CT perfusion cases varied widely among the five competing vendor workstations.

"Three-D workstations demonstrate a breadth of capability and variability of data," said Dr. Geoffrey Rubin, program director for the symposium and chief of cardiovascular imaging at Stanford University. "We need compelling, formal scientific evidence and clinical analysis of these automated tools."

OVERCOMING OBSTACLES

Barriers to widespread adoption of 3D reading remain. They include a lack of 3D integration with PACS workflow and tailored training to make the most of 3D features. Sites are also grappling with networking, storage, and archiving of the thousands of slices generated by CT devices.

Reimbursement also looms as an issue. While CPT codes exist for 3D postprocessing, payment favors technical, not professional, reimbursement, Barish said.

"We need to lobby to make sure that radiologists are reimbursed appropriately for the services provided," he said. "MPRs performed by a technologist on the CT scanner are one level of service, but it is another level entirely when a radiologist is generating advanced images such as multiplanar or oblique reconstructions. It should be paid at a level above and beyond the standard for routine sagittal and coronal images."

Where to install 3D workstations and how user privileges are granted is another area of concern. Should referring physicians have access to 3D tools? At Stanford, the answer is absolutely, Rubin said.

"We encourage referring physicians to interact with the images because it enables them to understand the data," he said. "I really don't feel they are going to compete with us for our jobs. That is not their interest."

Regardless of its value to diagnosis and display, if 3D is to be incorporated into the routine workday, workstations must be part of the routine. Workstations in the reading room are a must, Rubin said. Too many sites make the mistake of relegating 3D workstations to the CT suite or a specialized imaging lab. Accessibility is the key to regular use.

"The biggest impediment to mastering the software is having to go into another room, or even across the room, to use it," Teague said. "It needs to be where you are reading."

But perhaps the largest obstacle radiologists must overcome before they can cross the bridge to 3D use is their own training.

"The one thing we need to stop doing is something we've been doing our whole careers, and that's reading axial plane images," said Dr. Stuart Mirvis, director of emergency radiology at the University of Maryland in Baltimore. "We have to think three-dimensionally. There are pathologies that do not lend themselves to diagnosis when you limit yourself to the axial plane."