Three-D polishes ultrasound's image, pumps up demand

Newly developed ultrasound probe technology that acquires data in one sweeping stroke is propelling 3D ultrasound into new realms. Aided by software that reconstructs volumetric information, radiologists can review ultrasound data on workstations the same way they do CT and MR.

Volumetric imaging could amend ultrasound's operator-dependent reputation. And it could eliminate the need for time-consuming point-by-point exams, said Dr. Beryl Benacerraf, a clinical professor of radiology at Harvard Medical School.

"This is a new concept, not yet fully implemented," Benacerraf said. "But the idea is that the lengthy exam will not be needed anymore. The sweep of the anatomy with the probe will take about a minute, and then the data can be sent to a remote review area for analysis, as with CT or MR."

Several manufacturers have developed variations on volume rendering technology, but all use similar principles. The transducer tracks the ultrasound beam's coordinates as it sweeps an area of the anatomy. Software then reconstructs data for online or offline interpretation. Reconstruction can be performed as a surface rendering, already popular in fetal imaging, or as cross-sectional slices in multiple planes of view.

Such volumetric imaging, or "sonographic tomography" as Benacerraf calls it, could change current views held within radiology about ultrasound's diagnostic power and workflow. It could help the modality regain some respect-and radiologist practitioners.

Ultrasound has been a workhorse of diagnostic imaging for more than four decades. But in recent years, this reliable, safe, and inexpensive technology has lost ground to CT and MRI. The sexier imaging modalities have made serious inroads into traditional ultrasound-dominated areas and have taken disproportionately larger slices of the research funding pie. Several factors, including bleak reimbursement, explain ultrasound's dwindling appeal. But workflow looms large.

Patients complete a CT or MR scan and walk away shortly after the machines have saved all the relevant anatomy, to be reconstructed and analyzed later. Ultrasound does not afford that opportunity to sonologists and sonographers, who must painstakingly acquire data from a region of interest, one picture at a time.

Ultrasound also suffers from a lack of standardized scanning protocols, and it has a less than stellar record when it comes to reproducible study results, as compared with CT and MR. Though it won't happen overnight, volumetric imaging could address-if not solve-these problems.

Further improvements are needed in transducer technology, as are better reconstruction algorithms and easier ways to analyze the data. Dedicated technology now available, however, allows reliable quantitative assessments of volumetric ultrasound data as well as its rendering in multiple planes or in true 3D, said Flemming Forsberg, Ph.D., a professor of radiology at Thomas Jefferson University.

"Volumetric imaging will make ultrasound easier, more comfortable to perform, and less dependent on the operator," he said.

With a point-by-point examination using conventional ultrasound, operators may miss a particular anatomic detail if they can't see it or if the probe's scanning angle isn't right. A sonologist may question an examination's findings and order additional views. When the probe obtains a whole volume of data, however, the ability to "see" every little detail during scanning ceases to be an issue. All the information needed for interpretation is contained in the volume and in one take, Forsberg said.

SONOBREAKTHROUGHS

Three areas of volumetric imaging are fundamental to achieving ultrasound's promise in clinical practice: acquisition, image reconstruction, and interpretation. Researchers have made progress, with varied degrees of success, in all three.

In one study that tested an experimental volumetric imaging protocol, Benacerraf and colleagues scanned 25 second-trimester fetuses, using American Institute of Ultrasound in Medicine guidelines for five consecutive sweeps of fetal anatomy from head to toe.

The researchers reduced the time needed to perform and interpret a structural survey using a volume acquisition technique. It took them a mean time of 1.1 minutes to obtain the 3D volumes and 5.5 minutes to reconstruct the complete survey, half as long as a standard 2D survey. They noted the need for standardized acquisition protocols to minimize artifacts and ensure uniform images with the 3D techniques. But the group also confirmed that ultrasound can achieve workflow efficiencies similar to competing imaging modalities (Benacerraf et al. J Ultrasound Med 2005; 24:371-378).

Volumetric imaging gives practitioners the potential to provide more reproducible results, use more standardized and streamlined protocols, and complete the examination more quickly. It also allows reconstruction of images in different planes, no matter how the image was taken, making it less operator-dependent, Benacerraf said.

"This is the future of ultrasound," she said.

In a different study, Dr. Daniel A. Merton and colleagues at Thomas Jefferson University weighed conventional Doppler sonography's performance against that of a prototype system (Encore, VueSonix Sensors) capable of performing real-time volumetric imaging. They specifically assessed the time required for carotid artery blood flow studies as performed by an experienced and an inexperienced sonographer.

They found that sonographers, regardless of experience, could perform acquisitions on the 4D system almost three times faster than with conventional sonography. The time-efficiency difference they recorded was statistically significant (p = 0.0054). These tests need validation, however, in patients with carotid artery disease, Merton said. He presented the study at the AIUM meeting in June.

Issues regarding the time required for interpretation remain. Researchers have shown that it is possible to cut down the time required for the examination and for image reconstruction. But the technological breakthroughs have not extended their time-efficiency benefits into interpretation. Preliminary tests suggest that volumetric imaging may take longer to read, Forsberg said.

Sonographers and and sonologists, however, are just starting to discover volumetric imaging's possibilities. Most image processing computing power is now used to reduce noise and increase contrast. But sophisticated software algorithms already available, like those used by the military to interpret tiny details from satellite photography, or improved array technology could be applied into specialized volume processing techniques to make ultrasound examinations more cost-effective, said Dr. Brian Garra, clinical director of ultrasound at the University of Vermont College of Medicine.

"We are just getting into the image processing realm, but we have the computation power to improve accuracy. When a particular condition or disease that used to be hard to spot becomes obvious, you can make an interpretation faster. This will improve efficiency," Garra said.

Another study, presented at the AIUM meeting by Duke University researchers, reported preliminary findings with a prototype ultrasound array for high-speed volumetric imaging. The system features 512 channels that can transmit up to eight simultaneous beams of different frequency or phase. Using simulation software to measure phantom and in vivo scans, the researchers produced high-speed scans with large fields-of-view and an image quality similar to that of conventional 2D scans.

MANUFACTURERS' VISION

Several ultrasound manufacturers have seized on the opportunity to market individual solutions for volume acquisition, reconstruction, and visualization.

GE Healthcare launched its Volume Imaging Protocol (VIP) at the AIUM meeting. GE's VIP concept combines different technologies under its TruScan imaging platform, available with the Logiq scanner series. GE's imaging model allows for the acquisition of volumetric sweeps of raw data that can be reconstructed and analyzed in multiplanar displays or as true 3D in real-time or after the examination.

This can cut scanning time in half, said Terri Bresenham, GE vice president and general manager of global diagnostic ultrasound.

Dr. Mary Jane O'Neill, head of ultrasound at Massachusetts General Hospital, talked at the AIUM meeting about her preliminary experience with the VIP protocol. She observed that an average of 55 minutes elapse from the moment patients walk into the examination room until their release when conventional 2D ultrasound is used. The lion's share, about 30 minutes, is usually spent by sonographers scanning and saving images. Radiologists add 10 minutes for their reviews, but this time can increase significantly if rescanning is necessary.

Using the VIP protocol, the total examination takes about 25 minutes: 10 for sonographer scanning and 10 for review. Patients are no longer required to stay for rescanning, since the sonologist can reconstruct the information later, O'Neill said. More formal studies are needed to assess the efficacy and efficiency of this model, which has also been implemented at the Celebration Health/Florida Hospital System.

GE's imaging architecture enables the collection of data sets that allow sonographers to see the image on the screen while storing raw data in the background. With 2D ultrasound, rescanning was customary when imaging results were not satisfactory. VIP's raw data tool, on the other hand, can reprocess the information without the need for another examination, even from a remote location, Bresenham said.

"Many people think this model implies more work, but those who have tried it have discovered a significant improvement in productivity. Mass General researchers, who were the first to use these applications, have seen a backlog reduction. This sits well with administrators," she said.

Three-D pioneer Medison offers the 3D eXtended Imaging (XI) technology as the platform for its 3D MultiSlice Ultrasound concept, designed to perform scanning in a more uniform way than standard 2D ultrasound would allow, said Jeffrey Johnson, a marketing manager for Medison's ultrasound division. The package comprises three applications: MultiSlice View, Oblique View, and VolumeCT View.

The system provides radiologists with imaging reconstruction and display capabilities similar to those of CT or MR. The 3D ultrasound renderings can be seen as slices or as full volumes of anatomy. With a slice thickness down to 1 mm, the software allows multiplanar displays that fit varied imaging protocols, Johnson said.

"The slice thickness and multiplanar display make things more reproducible. This is going to be the next standard of 3D. It will impact not only ob/gyn, but radiology and other subspecialty imaging areas as well," he said.

Siemens has introduced fourSight 4D ultrasound software, available on its Sonoline products. The fourSight system allows real-time display and optimization of 3D and multiplanar images on the fly. Company sources say it eliminates the need to go back and forth between images during an ultrasound examination. Sonographers could optimize images with the touch of a button.

Philips Medical Systems' high-end ultrasound scanners have included volume imaging capabilities for many years. The xMATRIX technology recently made available on Philips' flagship iE33 cardiovascular ultrasound system and iU22 for radiology allows sonologists to explore several volume acquisition models, said Jim R. Brown, director of clinical and technical marketing.

"Before we jump on any kind of protocol-driven bandwagon, we need to investigate, with the aid of users, how we can leverage volumetric imaging to provide proven efficiencies in their imaging departments. There are many different user models, many different workflow scenarios, and you have to give the users the ability to adapt volumetric imaging to their specific workflow," Brown said.

Radiology departments across the U.S. will be looking at how they can increase their workflow efficiencies in ultrasound and make them less technically dependent and more valuable to diagnosis within the next two to four years. Radiologists will probably need several more years after they and their labs start adopting different types of workflow scenarios. Smaller community hospitals won't see these advances until they are fully implemented in the major institutions, where PACS and DICOM protocols for ultrasound are not yet the norm, Brown said.

Still, industry officials predict that volumetric imaging could ignite ultrasound's revival.

"We are on the verge of a renaissance for ultrasound," Bresenham said. "The same transformation that occurred in CT and MR five years ago will happen to ultrasound."

CLINICAL IMPACT

One potential drag on volume imaging's momentum is the fact that successful applications of the technique in ob/gyn ultrasound may not carry over to every clinical subspecialty. Advocates say that it's all a matter of technique and expertise.

"Ultrasound is a game of acoustic windows. The good sonographer, or anybody who does good ultrasound, knows how to get those acoustic windows," Garra said.

But not all imaging areas have an acoustic window as good as that of fetal applications. Some organs, like the breasts, are amenable to volumetric imaging because they have appropriate acoustic windows. In the abdomen, however, good acoustic windows large enough for volume scanning can be difficult to find.

"The window you are using for optimal coverage in kidney imaging, for instance, is not going to give you a great image of the liver. And for the liver, you've got a problem with the ribs," he said.

There are ways around these problems. Coverage for the liver, for example, could be performed by sweeping each lobe separately or in quadrants. As with multislice CT, radiologists don't need to reconstruct the whole organ. Thin MSCT acquisitions allow reconstruction of an area of interest. The same can be accomplished using volumetric ultrasound, Garra said.

Available clinical data show this is possible in several imaging areas. Dr. Marilyn A. Roubidoux and colleagues reported on 34 women who underwent volumetric ultrasound and color Doppler to evaluate treatment for early-stage breast cancer. They acquired 0.5-mm slices in gray-scale and color Doppler mode, reconstructed as volumes on a separate workstation. As shown in the previous studies, the 3D display was useful for the quantitative analysis of breast malignancy's vascularity (J Ultrasound Med 2005;24:885-895). Preliminary studies in such diverse areas as cardiovascular, genitourinary, neurologic, and pediatric imaging also show promising results.

In terms of workflow, however, the benefit of volumetric ultrasound transcends the clinical realm. Several studies published by European and North American researchers provide evidence on the physical toll exacted on sonographers. Musculoskeletal injuries sustained by sonographers have huge economic implications, including absenteeism and permanent disability. A survey of Canadian sonographers documented significant associations between their work and musculoskeletal conditions. Though only a few were absent from work due to their symptoms, more than two-thirds of the respondents reported they worked in pain (Appl Ergon 2002; 33[5]:385-393). Volumetric imaging could help in the evolution of the sonographer's role.

"Many of these injuries are career-stopping. Sonographers would welcome the opportunity to vary their days, so they could spend some time taking pictures and dedicating other time reconstructing the volumes in such a way that radiologists can see them quickly," Benacerraf said.

Mr. Abella is an assistant editor of Diagnostic Imaging.