Karen Sandrick

The hand-carried ultrasound market over the last few years has been dominated by three companies. Early leaders SonoSite, GE, and Zonare accounted for more than 85% of the overall revenues generated by sales of compact sonographic units in 2007.

Well, move over, Big Three, as two mainstream radiology players enter the field. Philips and Toshiba are marketing high-end general-purpose hand-carried products that can perform routine radiology and cardiology scans in virtually any setting. Philips' CX50 and Toshiba's Viamo are laptop-sized ultrasound systems that were launched in Europe in March and will be displayed for the U.S. audience at the upcoming RSNA meeting.

The entry of the two major manufacturers speaks to the steady growth of the hand-carried ultrasound sector. Global revenues from compact sonography systems grew by 42% to $565 million in 2007, according to the second annual hand-carried ultrasound report by Klein Biomedical Consultants.

Future potential is huge. The forecast is that compact ultrasound will become a billion-dollar market in just the 10 years, said Jeff Green, senior manager for ultrasound, Toshiba American Medical Systems.

Technological advancements are making portable ultrasound more widely accepted by clinicians. Initial compact units were fine for triage or image-guided procedures, but they didn't rise to the diagnostic level, said Jim Brown, marketing and product manager for Philips portable ultrasound systems.

Improved, diagnostic-quality images, longer battery life, and better tissue penetration are attracting increasing numbers of customers.

"We've improved over the years so overall image quality on hand-carried units is about the same as that of the console," said Al Lojweski, global general manager for GE's cardiovascular ultrasound division. "The advanced features-the 4D-may not be there, but basic 2D, B mode, color Doppler technology in compact systems is equivalent to a system you'd find in the ultrasound lab."

Speed and ease of use are also factors. Physicians who in the past, would have had to wait for the results of a scan or learn how to operate sophisticated machinery can now diagnose patients practically as they come in the door.

"A physician can do an ultrasound to look for free fluid in the abdomen and make a decision in two minutes without having to wait for an ultrasound tech to get there," said Dave Willis, vice president of competitive strategy and innovation at SonoSite. "Algorithms automatically make changes, which speeds up exam time and gives good images so physicians can move from one patient to the next without figuring out a bunch of controls."

As a result, almost every clinical department is using portable ultrasound, including anesthesiology for nerve blocks, radiology, cardiology, intensive and critical care, and the emergency department. SonoSite, in fact, sells to 20 different medical specialties, said Anne M. Bugge, vice president of corporate communications and demand generation for SonoSite.

A flurry of new releases this year is offering clinicians more choices. Zonare is now shipping the z.one ultra sp, a full-featured, cart-based system that can be converted into a premium compact system.

SonoSite is selling the latest addition to its hand-carried product line internationally and is awaiting FDA approval for U.S. sales. The NanoMaxx is a six-pound point-of-care system based on SonoSite's fourth-generation Turbo technology and offers advanced algorithms as well as a fluid-resistant user interface.

The 2.0 release of Siemens' portable Acuson P50 is providing electrophysiologists and interventional cardiologists with real-time, diagnostic quality images as well as Doppler blood flow information. The new release also has 256 beam-forming channels for improved imaging quality as well as SieClear multiview spatial compounding and SieVision speckle reduction.

GE's Venue 40, first shown in May and formally unveiled in September, is a flat-screen, tablet-type hand-carried ultrasound unit. It has a base for docking or can be mounted on a wall or the ceiling. The Venue 40 can be used for guiding vascular access, needle biopsies, or interventional ablations as well as examining patients in critical or intensive care units or the emergency department.

The undeniable draw of portability has been married to top-notch image performance.

"Miniaturizing the ultrasound console at first made it much easier for sonographers and physicians to do exams in places where space was limited-physicians' offices, ER bays, or the OR," Lojweski said. "Giving performance and portability is stimulating a new interest in the marketplace."

Articles

Miniaturization gives portable units functions once available only in cart-based systems.

Mainstream manufacturers embrace hand-carried ultrasound

Image-guided intensity-modulated radiotherapy, high-intensity focused ultrasound, and cryotherapy are increasing the curative treatment options for men with prostate cancer. The problem is how to determine which patients are most suitable for these therapies.

The new treatments for prostate cancer require precise localization of disease, but routinely used imaging techniques are not sensitive or specific enough for accurate staging. Gray-scale transrectal ultrasound has a sensitivity of 30% to 50% and specificity of 77% to 90% for local staging and tumor localization, and 1.5T MRI has a joint sensitivity and specificity of 71% to 74% for conventional anatomical prostate imaging, according to recent research.

High-field MRI could be the answer. Attendees at an ECR 2009 special focus session on prostate cancer will learn about the advantages of 3T MRI for prostate cancer imaging. 





"MR is a powerful method for imaging the prostate gland because its high spatial resolution provides excellent soft-tissue contrast, which may aid the detection and localization of malignant lesions, the evaluation of disease stage based on the extraprostatic extension of cancer, and the detection of local or distant disease recurrence after treatment," said Prof. Dr. Jurgen J. Fütterer, a professor of radiology at Radboud University Nijmegen Medical Centre in the Netherlands.

He will describe what 3T MRI adds to prostate imaging.

The signal-to-noise ratio increases almost linearly with magnetic field strength, whereas the noise remains nearly unchanged with 3T MRI. Thus, the SNR is approximately two times higher at 3T compared with 1.5T. The most significant benefit of this difference is the increase of spatial resolution, which improves the depiction of anatomical details and/or shortens acquisition time, thereby increasing patient throughput.

The chemical shift effect also increases linearly with magnetic field strength. Spectral resolution at 3T is therefore enhanced compared with 1.5T. Signal intensity and frequency dispersion are higher, which may lead to better tumor characterization.

In dynamic contrast-enhanced MR imaging, the increase in SNR can be used to improve the temporal resolution of dynamic measurements. Increased temporal resolution may improve the accuracy of measurement of pharmacokinetic parameters. 

Endorectal 3T T2-weighted and high spatial resolution dynamic contrast-enhanced MRI are able to identify morphological and vascular details, which may guide not only the detection and staging of disease but the direction of biopsies and treatment. Signal intensity changes obtained during dynamic image acquisition provide an estimate of the amount of contrast material that has accumulated in lesions. 

The passage of contrast into and out of lesions reflects the kinetic properties of tissue physiology, such as K

 and Ve values, and provides an indication of microvascular permeability and angiogenetic potential, according to Dr. B. Nicholas Bloch, an instructor in radiology at Beth Israel Deaconess Medical Center in Boston. He will discuss MR perfusion and high-resolution imaging. 

The high field strength of 3T MRI is particularly useful in dynamic contrast-enhanced studies of the prostate gland.

"Dynamic contrast-enhanced MRI makes it possible to combine fast imaging -- less than three seconds per image set -- with good spatial resolution. This technique directly benefits from the increased SNR at 3T," Fütterer said.

High-field MRI has its drawbacks and challenges, however, including a quadrupling of radiofrequency power deposition compared with 1.5T, shorter T2- and longer T1-relaxation times, increased susceptibility differences, dielectric effect, and signal heterogeneity from larger B1 field variations. Using thinner slices and increasing the spatial resolution or bandwidth at higher field may partially circumvent some of these issues, but these changes will offset some of the SNR increase gained with 3T imaging. 

In addition, the loss in SNR is less than linear with voxel size due to the decreasing line width (e.g., when resolution is increased from 0.075 to 0.094 cm

, the SNR decreases between 44% and 60%). Due to radiofrequency penetration effects, it can be difficult to achieve good RF field homogeneity. Parallel imaging techniques, increased TR, reduced flip angle, and increased RF pulse duration can be used to overcome these problems, but at the expense of SNR.

Despite these difficulties, the increase in SNR provided by 3T MRI is likely to expand the potential clinical applications for evaluating the prostate, Fütterer said.

"Endorectal 1.5T MRI is a common method of evaluating men with suspected prostate cancer. But in the next few years, 3T MRI will become the standard. An SNR at 3T that is equal to the SNR at 1.5T with an endorectal coil is a big advantage," he said.

Image-guided intensity-modulated radiotherapy, high-intensity focused ultrasound, and cryotherapy are increasing the curative treatment options for men with prostate cancer. The problem is how to determine which patients are most suitable for these therapies.

Evidence grows for high-field MRI’s value in prostate cancer

Conventional radiography/fluoroscopy has always been cumbersome.

The equipment is big and bulky, and the examination is time-consuming, as technologists wrestle with cassettes between exposures and wait for film to develop.

The introduction of digital flatpanel detectors is not only streamlining the configuration of R/F rooms and simplifying throughput. It is also improving the detection of pathology, reducing radiation dose, and increasing patient comfort. It is assessing motion in real-time as well, so radiotherapy can be delivered in the highest tolerable dose to the most precisely defined tumor target.

Flat-panel detectors from major suppliers such as Varian and Trixell can be found in R/F systems by GE, Hitachi, Philips, Shimadzu, Siemens, and Toshiba.

One advantage of the digital flat panel is its large area. In contrast to image intensifiers, which have a maximum size of 16 inches, general-purpose flat-panel detectors are 14 x 17 or 17 x 17 inches. The larger coverage means fewer exposures and faster exam time.

“Doctors can catch live motion video, take exposures here and there, and do all of the spot filming they want on one exposure. Generally in the past, techs would have to step in and do a series of five or six exposures.

Imaging has gotten much faster because there is no cassette handling. Now the radiologist pushes a button, the exposure is made, and a few seconds later, the radiologist gets the image and can go on to the next. He or she doesn't have to switch between the radiography and fluoroscopy world,” said Tom Truesdell, manager of radiography/fluoroscopy for Siemens Medical Solutions. Speed means not only better throughput; it also improves lesion detection.

“The response time of a digital flatpanel R/F system is very fast. You are on rapid fire taking images. So you are more likely to capture the pathology you are seeking. You also know right away whether you have the image you want,” said Dr. Edward Lubat, managing partner of Radiology Associates of Ridgewood in Waldwick, NJ.

Unlike image intensifiers, which are not geometrically accurate and produce pincushion distortion at the edges, flat panels have uniform image quality from edge to edge. Flat-panel technology also has higher detective quantum efficiency (DQE) than image intensifier charge-coupled device camera systems, which translates into superior small area contrast resolution and improved visibility of small vessels and catheters, according to Dr. Rick Colbeth, vice president of flatpanel technology for Varian.

Flat-panel technology has greatly improved image quality on the fluoroscopy side. As a result, radiologists in Europe are finding that they can use lower-dose fluoroscopy and still capture diagnostic events without needing to switch to high-dose, highresolution radiography. Prof. Yvan Palmers, director of imaging services for Ziekenhuis Oost-Limburg in Ghent, Belgium, has acquired a diagnostic- quality image in the fluoroscopy mode that used 30% of the radiation dose required by the traditional radiography mode.

Flat panels are smaller and lighter than image intensifier CCD camera systems, so digital R/F systems tend to be more maneuverable. Added space beneath the system allows the table to be lowered to a 19-inch height for easy and more comfortable patient transfers. Reduced ancillary equipment and increased distance between the table and the imaging tower facilitate patient positioning. If it weren't for flat-panel technology, real-time gating during imageguided radiation therapy would be impossible, said Dr. Todd Scarbrough, medical director of Melbourne Internal Medical Associates Cancer Center in Melbourne, FL.

Now that a flat-panel device has been attached to a linear accelerator, Scarbrough can track the movement of a fiducial metallic marker on the surface of a tumor as it moves with the respiratory cycle in order to synchronize the release of a beam of radiation.

“We can't move the radiation beam around, but we can find out where the tumor is and turn the beam on or off depending on the respiratory cycle. It's important for us to be exceedingly accurate with how we delivery radiotherapy, and flat-panel imaging lets us do that. It allows us to decrease side effects and go to a higher dose because we are confident about where the radiation is going,” Scarbrough said.

Conventional radiography/fluoroscopyhas always been cumbersome.

Flat-panel detection speedsdiagnosis and throughput

Since full-field mammography started going digital eight years ago, systems based on flat-panel arrays have dominated. But computed radiography has begun seeping into that market and, if the FDA allows CR companies to apply for streamlined 510(k) approval of mammography upgrades, the trickle of CR products could turn into a flood.

 With the appearance of CR-based full-field digital mammography products, mammographers can choose whether to ease into digital by extending the life cycle of already installed equipment or invest in flat-panel systems as the bridge to a 3D future.

Until lately, flat-panel mammography systems have been the only game in town. A little over 40% of the 13,000-plus Mammography Quality Standardization Act-accredited facilities currently have FFDM systems, and most are built around solid-state electronics, sometimes referred to as being based in digital radiography.

 The only company to have an FDAapproved CR mammography system, Fujifilm Medical Systems USA, has sold 500 Fuji CR for mammography (FCRm) units in the U.S. since the product was launched in 2006. Flat-panel systems have been popular because they are highly efficient. They allow imaging centers to get more patients through screening in less time. The Mammomat Inspiration Siemens Medical Solutions requires only 27 seconds between exposure times and 18 seconds between the exposure and display of digital images. The Inspiration can therefore examine more than 15 patients an hour.

 CR mammography has trouble competing in terms of throughput. "With CR, you still have the interaction with the technologist, as far as handling cassettes," said Pat Montgomery, manager of women's health for Agfa HealthCare, a maker of CR products. "The cassettes are digital as opposed to having film , but the technologist still has to put a cassette into a CR reader, wait for processing, then take it out. If a mammography site has high volume or a patient backlog [and] would benefit by getting more patients through, it probably would want to consider DR."

 DR systems also lend themselves to advanced applications. They bring high image quality at low doses, so the technology has the potential to obtain sequenced images that can be combined and reconstructed into a 3D data set. CR does not have that potential, said Jose Martinez Abellan, marketing manager for GE Healthcare's mammography business, a maker of flat-panel FFDM systems. A downside of DR is cost.

 Although the price tag for DR-based systems has declined steadily since 2000, it may not be enough. DR manufacturers are beginning to offer economical versions of their products. Hologic's Selenia S, which is targeted at freestanding clinics and the mobile market, has fewer bells and whistles than the full-blown Selenia system and is more competitively priced against CR systems than full-feature solid-state FFDM products, yet the price differential remains an issue.

 Flat-panel DR-based FFDM products typically cost around $350,000, while CR-based units run between $140,000 and $240,000. So while some facilities may opt for DR in their principal screening center, they could find CR more appealing, price-wise, for remote locations, said Stephen Archer, director of worldwide marketing for mammography solutions at Carestream Health, which-like Agfa-is developing a CR-based FFDM for sale in the U.S.

Manufacturers, such as Carestream, Fuji, and Agfa, that sell CR mammography products outside the U.S. appear poised to effect a swift introduction of their products into this country, now that the FDA is actively considering downgrading CR mammography to a class of device that requires less rigorous approval. The FDA closed the public comment period concerning a possible switch from premarket approval to 510(k) status for CR mammography at the end of August. A decision on when and how to reclassify FFDM is expected in the coming months.

 If the FDA does make the change, which may not happen until late winter or early spring, CR companies will be able to sell in the U.S. the exact same mammography upgrades they market elsewhere. Agfa sells mammography add-ons to its CR 85-X multiplate and CR 35-X singleplate readers throughout the world. Carestream Health offers the DirectView CR 975, which can process up to eight plates at once, and two single-plate options: the Elite and the Classic, which differ in speed and price. And Konica Minolta markets the Regius Xpress CR.

 CR manufacturers base their hopes of gaining a wider presence in mammography not only on their lower price points. They also stress that CR offers an easy digital conversion.

 "There is no disruption of workflow going from film to CR," said Jeanine Rader, director of women's healthcare for Fujifilm. "Technologists do the same thing they did before.

 And there is no need to take walls down or make changes to accommodate new flat-panel technology."

 CR also fits right in with existing analog technology.

 "If a facility has been doing screen-film mammography and they have an x-ray generator that is reasonably modern or that they don't want to replace, maybe because of cost, CR works very well," Archer said.

Despite the differences, DR and CR have one challenge in common: how to manage images from other modalities and different PACS and RIS vendors.

 GE Healthcare has obtained FDA approval to sell a mammography workflow solution from Image Diagnost International, a German software company it recently acquired. The product, which will be formally launched at the RSNA meeting, can be used by imaging centers of all sizes that do not have a high volume of screening studies but nevertheless have workflow problems and by academic medical centers that conduct research on tomosynthesis or other imaging devices.

 "The company has developed workflow solutions in screening," Abellan said. "It has built all the interfaces for use with multiple PACS and RIS vendors. It has done a huge amount of work on how to relate previous images to current images and on how to do all the work, from entering the patient into the HIS/RIS to doing the diagnosis on the workstation."

 Konica Minolta has developed a distributed, rather than centralized, workflow plan around the CR-based system the company markets outside the U.S. and hopes to bring it to this country in the months ahead. It eliminates the need to move from the exam room to the central processor and wait for processing.

 Agfa and Fuji are offering vendor-neutral multimodality diagnostic review stations. Fuji's Synapse 3.2.1 workstation is not limited by modality. It not only handles the screen-ing images that come from digital detectors, but also displays MR, ultrasound, and CT images of the breast.

 Agfa's Impax MA3000 diagnostic workstation links digital mammography with other forms of breast imaging as well as chest x-rays, PET scans, and any other images on a PACS network.

 "Radiologists can sit down and do all of their work from one location versus having to move from workstation to workstation," Montgomery said. "This will translate into benefits for patient care because the point is to make the radiologist more efficient."

A nagging question for the evolving digital mammography market involves tomosynthesis, a technology dependent on flat-panel detectors that deliver volumetric images of the breast that cannot be generated with CR. Tomosynthesis promises greater sensitivity and specificity in detecting breast cancers at lower doses and with less compression.

 "There is the possibility that one tomo view would replace two regular views on mammography, and that means half the dose of a regular mammogram," said Jonny Eser, product manager for women's health at Siemens Medical Solutions. "And by slicing through the breast with tomosynthesis, you should be able to see lesions without compressing the breast to find masses that hide under the glandular tissue."

 The relevance of tomosynthesis depends, however, on its commercialization. Hologic, the only firm so far to submit data to the FDA for approval of a tomosynthesis unit, has been fielding requests from reviewers for clarification of the data. The company nevertheless expects to begin marketing the product at the beginning of 2009. Hologic has already launched tomosynthesis in Europe and will begin shipping units in December, according to CEO Jack Cumming.

 Even after its entry into the U.S., widespread adoption of tomosynthesis will take time. The technology most likely will be used mostly in research settings and by academic medical centers initially, until the procedure is widely reimbursed. Whether and when this happens may depend on research to establish if tomosynthesis will complement or replace fullfield mammography. The practical use of the technology also depends on vendors finding ways to manage its volumetric images. Tomosynthesis developers are testing prototype workstations.

 "With tomosynthesis, you have to address 30 times the quantity of data you get today, and time is critical," said David Caumartin, general manager of global mammography for GE Healthcare. "If you need more time to review the tomosynthesis images, then you need more radiologists, which we will not get. So we are working on ways to handle this huge amount of information."

Since full-field mammographystarted going digital eight yearsago, systems based on flat-panelarrays have dominated. But computedradiography has begun seeping intothat market and, if the FDA allowsCR companies to apply for streamlined510(k) approval of mammographyupgrades, the trickle of CR productscould turn into a flood.

CR prepares to challenge solid-state digital mammo

It's a given that breast cancer screening reduces mortality from the second leading cause of death of women in the U.S. Early detection of breast cancer requires screening techniques that can reliably detect lesions in the 5 to 15-mm range. Screening techniques have difficulty finding abnormalities of this size, however. Conventional scintimammography often has an abysmal sensitivity, said Dr. Michael O'Connor, a radiologist from Mayo Clinic in Rochester, MN. In a 2002 study of 50 women, general-purpose scintimammography had a sensitivity of 35% to 64% for lesions less than 1 cm, O'Connor reported at the 2008 SNM meeting in June (J Nucl Med 2002;43(7):909-915).

Ultrasound adds little. Ultrasound and mammography had a sensitivity of 50% in a study of 2500 women who were at high risk for breast cancer, O'Connor said (JAMA 2008;299[18];2151-2163).

While the sensitivity of MRI is impressive-around 80%-its specificity can be as low as 50%, and it is 10 times more costly than mammography, O'Connor said at the SNM meeting.

So what's the answer? Enter nuclear medicine.

Early detection of breast cancer requires imaging techniques to reliably find lesions of 5 to 15 mm. Conventional scintimammography in the 1990s admittedly was dreadful at detecting abnormalities in this range. Its sensitivity was 35% to 64% for lesions smaller than 1 cm.

But newer versions of nuclear medicine devices that are dedicated to breast imaging are finding extremely small tumors. As reported at the 2008 SNM meeting, a prototype gamma camera-based breast imaging system developed at the Mayo Clinic over the last six years has found 10 cancers in 762 high-risk patients who were monitored with the device over time. Six of the tumors were 10 mm or smaller, and one was 15 mm. Mammography, meanwhile, found only three cancers of any size.

A new PET system designed for imaging small body parts, such as the breast, has a resolution of 1.5 mm, which is about the size of the breast's network of ducts where most breast cancers arise, said Dr. Kathy Schilling, medical director of the Center for Breast Care in Boca Raton, FL. In a study of 136 patients, the PEM Flex Solo II (Naviscan PET Systems, San Diego) identified a ductal carcinoma in situ only 2 mm in size that was missed on MRI.

An experimental PET/CT device that is under development at the University of California, Davis achieves a resolution of 300 microns for CT and 2 mm for PET, which allowed investigators in the first human trial with the device to find a hot spot in the breast that was about 5 mm across. By comparison, whole-body PET/CT scans would not have been able to find a lesion smaller than 1 cm in the breast, said Dr. Ramsey D. Badawi, an assistant professor of radiology at UC Davis.

These dedicated breast imaging devices just may address the shortcomings of mammography and MRI. The breast imaging system from Mayo Clinic was significantly more sensitive than mammography in the study reported at the SNM meeting. In a 15-month follow-up period, the system identified six of seven cancers in 341 patients for a sensitivity of 85.7%. Mammography detected only two of the seven for a sensitivity of 28.6%.

The system also had a lower recall rate than mammography. Fifty-six of 762 patients were recalled for further investigation after imaging with the gamma camera (7.3%) compared with 70 (9.2%) after mammography. The biopsy rate was higher (3.4% versus 2%), and so was the positive yield at biopsy (28.6% versus 20%) for the gamma camera, O'Connor said.

The PEM system was more sensitive in detecting DCIS than MRI (91% versus 83%), and its performance was not affected by circulating hormones."MRI has a tough time with patients if they have natural circulating hormones or are on hormone replacement therapy. There are a lot of false-positive results with MRI. We do not see that same effect with PEM. Hormones don't cause things to light up that are natural in women who are in the premenopausal or per-menopausal state," Schilling said.

Nuclear medicine devices are less costly than MR scanner. Mayo's gamma camera is estimated to be one-fifth to one-third the cost of a breast MR, according to O'Connor. 

The cost of the PEM machine is about half that of a whole-body MR scanner, Schilling said.

And nuclear medicine systems don't require a major facility or workflow overhaul. The Mayo and UC Davis prototypes are still under development, but the PEM is in use around the country. It is easier to get up and running than MR, according to Schilling.

"It's very difficult to set up a breast MRI program. Physicians around the country struggle with it, because there are no standardized protocols for image acquisition or even interpretation. That's not so with dedicated PET breast imaging," Schilling said. "PEM is smaller than a mammography unit. You just plug it in the wall and go."

Operators must have a nuclear license, but there isn't all the room preparation and shielding that are needed for MRI, she said. "And for image interpretation, there's a very short learning curve. So it's easy to implement in a breast practice," Schilling said. 

The Mayo prototype is a dual-detector system that uses cadmium zinctelluride technology. The gamma camera is ideal for breast imaging because it achieves a high degree of resolution. Intrinsic resolution is 1.6 mm, and energy resolution is 4%. It also has minimal dead space. It may be preferable to mammography because it requires less compression. The compression force is about one-third that of mammography.

"A big drawback, however, is radiation dose," O'Connor said.

Patients are injected with 740 MBq of technetium-99m sestamibi five minutes before being imaged with the gamma camera. The 6.7-mSv total dose of radiation for gamma camera imaging is therefore markedly higher than the minimal 1 mSv or less for mammography. But investigators at Mayo Clinic are working on techniques that will lower the dose of sestamibi to 70 to 100 MBq.

The PEM system is a commercially available high-resolution functional tool for breast imaging. Unlike MRI, which identifies cancers by looking for new blood vessel formation, the PEM looks for differences in metabolic uptake of sugar in normal and abnormal cells. And unlike whole-body PET, for which high detection rates of breast cancers are hampered by the arrangement of the detectors and their distance from the target organ, the PEM can see cancers when they are still confined to ductal structures because its detectors are in intimate contact with the breast, Schilling said.

The experimental dedicated PET/ CT breast imaging machine at UC Davis is capable of high-resolution anatomic as well as functional imaging because it is not imaging the chest, where both PET and CT signals are badly attenuated and signal is lost during whole-body imaging. The scanner consists of flat-panel detectors, which have been used for high-resolution PET/CT imaging before. But it also includes very small scintillating crystal arrays, which were pioneered in small animal imaging and have not been used in humans until now.

In its first test in humans, the PET/CT scanner found three areas of ductal carcinoma in situ, including one with intralymphatic cancer in a 49-year-old woman who had pathologically confirmed breast cancer.

Findings from the studies presented at the SNM meeting are carving out several paths for dedicated nuclear medicine breast imaging.Because of its high sensitivity and low cost, Mayo's gamma camera may be an alternative to MRI for screening women with a personal or family history of breast cancer or mutated BRCA1 or BRCA 2 genes. The technique is two times more sensitive than mammography and just as sensitive as MRI in finding breast cancers among high-risk women, O'Connor said.

Breast imaging with the gamma camera also has a higher specificity than mammography or MRI. In the study reported at the SNM meeting, the technique had a specificity of nearly 94% while mammography had a specificity of 91%. The specificity in this study bested that of MRI, which, in the best of circumstances, tops out at around 80%.

If the PEM proves to compare well in a direct one-on-one match up with MRI, it may replace MRI in small, cost-conscious breast imaging centers. Schilling is currently leading a multicenter trial of the PEM and MRI in 400 women with newly diagnosed breast cancer. The results of that trial, which are expected early next year, may open the door to more wide-scale adoption of the technique in women's health practices that can't afford or don't have access to MRI.

"Instead of just waiting until tumors get so big we can see them and feel them, we're trying to identify the earliest changes down at the cellular level and the changes that make cancer cells different from normal cells. We are looking for a tool that will not be redundant and find the same cancers as breast MRI," Schilling said. "We are able to find additional cancers and even precancerous pathology with PEM, with improvement in our specificity and accuracy."

In a comprehensive breast center, she said, an exam with the PEM would complement mammography.

"It would be one of multiple tools that you can tailor for specific patients. In a smaller center, or one that does not have breast MRI, this can serve as a functional imaging tool to look for breast cancer in high-risk women." Schilling said.

Dedicated breast PET/CT is not for everyone. It takes more than an hour for the radioactive tracer necessary for the PET portion of the exam to circulate into the breast. PET/CT, therefore, is probably not going to be practical for screening large numbers of women, Badawi said.But it may have a role if a mammogram is suspicious for breast cancer and a woman is destined for biopsy.

"Our technology may be able to reduce the number of negative biopsies that these women have. If we can see lesions at a high resolution and determine if they are cancer or scar tissue-or even a cyst-then there would be no need to do the biopsy," Badawi said.PET/CT may also be helpful in surgical planning to depict the full extent of disease so the surgeon has a better chance of removing all of the involved areas at the first surgery.

Another emerging role for breast PET/CT is as a chemotherapy treatment monitor.

"There has been a lot of interest in giving chemotherapy before surgery to shrink the tumor so it is small enough to cut out completely. But how do we know which chemotherapy is going to work?" Badawi said.

It may take several months of monitoring with CT to see whether the tumor shrinks.

"But with PET scans, we can see that the uptake of sugar goes down quite quickly if the chemotherapy is working-even in a day or two. On that basis, you could give the drug, do the PET/CT scan, and see if the cancer is changing or responding. Then you have the answer right away," Badawi said.

The new breast imaging techniques are giving clinicians and radiologists more options for deciding how to image the patient, not just the disease.

"Medicine is becoming much more personalized today, not only in treatment but also in diagnosis," Schilling said. "So we are really looking at the patient and her age, breast density, and risk for developing cancer, and we are beginning to tailor the screening to fit her needs." 

Ms. Sandrick is a freelance writer and frequent contributor to Diagnostic Imaging. 

Nuclear medicine tackles limits of other screening options

Computer-aided detection is evolving from an interesting technological trapping to a standard of care. Part of that process involves the ongoing iterative advancement expected to be on display at the RSNA meeting. 

While CAD breast imaging products are being revamped to provide gains in clinical accuracy and utility, CT-oriented product vendors are focusing on high-end visualization and advanced analyses to increase productivity as well as accuracy. 

The first CAD mammography products, with detection rates in the high 80% range, did not command much respect. More recent CAD products, however, have raised detection rates to the mid- to upper 90% range, even for difficult-to-spot signs of breast cancers such as microcalcifications and invasive lobular carcinomas. 

CAD mammography identified 98% of the microcalcifications and 95% of the invasive lobular carcinomas in a study of 273 women who had histologically confirmed breast cancer, and it spotted 89% of the mixed mass/microcalcification lesions as well as 84% of the masses. The study, which used the Second Look CAD unit from iCAD, was conducted at George Washington University Medical Center. 

The iCAD family of Second Look products for screening and diagnostic mammography incorporates an enhanced CAD system, which has an overall sensitivity up to 94% and specifically targets the identification of microcalcifications. 

"Calcifications are on the edge of what you can see, so providing a set of 'eyes' that is working at the depth the image is taken gives the radiologist the ability to see much finer detail," said Janet Sterrit, vice president of the mammography business unit for iCAD. 

Empirical data and trends analysis of the effectiveness of mammography CAD have been so convincing that Aetna revamped its clinical policies in April to include CAD as an adjunct to mammography. Aetna had previously declined to reimburse for breast cancer screening mammography with CAD. 

Several companies have developed workstations with integrated mammography CAD systems. At the RSNA meeting, iCAD will display Second Look 700, a film-to-digital CAD for screening and diagnostic mammography, featuring a built-in monitor. The workstation, designed for use in centers with heavy patient loads, offers an easy-to-use workflow option that is driven by patient lists. The company will also show Second Look Digital, which is compatible with digital systems from GE Healthcare, Hologic, IMS Giotto, and Siemens Medical Solutions. The digital product automatically receives and sends results via a DICOM network without requiring manual interaction, Sterrit said. 

CAD for breast MRI is not only proving itself clinically, it is boosting productivity. In papers presented at the 2003 RSNA meeting and the 2005 European Congress of Radiology, CAD for breast MRI achieved 100% sensitivity for detecting enhancement kinetics in malignant breast tumors, 94% correlation with histologic analysis of breast tissue, and a reduction in artifacts during subtraction. 

In addition to a study presented at last year's RSNA meeting, which found that CAD for breast MRI could decrease radiologists' time by 80% and technologists' time by 90%, anecdotal reports document dramatic improvements in workflow. 

Dr. Nanette Debruhl, an assistant professor of radiology at UCLA Medical Center, used to spend hours interpreting MRI breast studies. While a patient was being imaged, Debruhl would sit with the technologist and decide which calculations to make. Processing and interpreting a single study could take an average of 45 to 60 minutes. 

With CADstream, Confirma's CAD system for MR breast imaging, Debruhl can obtain the information she needs to make a diagnosis in less than 15 minutes. All the technologist must do is send the scans to CADstream's Internet address and wait for the postprocessing. Because of CAD, Debruhl's department averages 10 cases a day. 

"CADstream allows us to evaluate areas of the breast that take up more contrast relative to the surrounding tissue and magnify the images to look at the borders of a mass clearly," she said. "It has markedly improved the speed of analysis, and with the number of films and studies that are being done now-we do an average of 2000 images per case-speed is important to free up the technologist's time as well as the radiologist's," she said. 

Confirma will demonstrate at the RSNA meeting how CAD may play a role in interventional planning and guidance for breast cancer. The vendor will display its latest version of CADstream, as well as the Access Breast Coil, which was shown as a prototype last year and is expected to be approved by the FDA for use with the Siemens 1.5 T MRI by meeting time. 

Riverain Medical will make its second appearance at the RSNA meeting with its chest radiography CAD product. Although most lung CAD systems use CT images, there is an important place for CAD in x-ray, which is the first-line imaging strategy for lung cancer, according to Kevin McBride, vice president for marketing. The Riverain system offers both lateral and PA images and can be integrated with a PACS. 

Vital Images is showcasing its integrated CAD products. In addition to R2 Technology's lung CT CAD package, the company will exhibit Confirma's CADstream for breast imaging. It also plans to launch a product that integrates the computer-assisted reading software for colon polyps from Medicsight, which has expanded its colon algorithms since it received clearance from the FDA in the fall of 2004. The company is presenting seven abstracts attesting to the performance of its algorithms. 

The integrated CAD products are designed to simplify radiologists' daily routine. Rather than having to move from one workstation to another to do advanced analysis and turn on CAD, they can perform both operations on one workstation. 

"While radiologists are doing their advanced analysis for any number of applications, they can get their CAD results within the same workflow, without physically going to another place," said Susan A. Wood, executive vice president of marketing and clinical development at Vital Images. "The strategy is to get CAD results outbound. If you take that to the next level, the advanced analysis component is being integrated into the PACS solution, so PACS companies that traditionally don't have this capacity can work together with CAD to optimize workflow for the radiologist." 

Karen Sandrick, a contributing editor for Diagnostic Imaging. 

Computer-aided detection is evolving from an interesting technological trapping to a standard of care. Part of that process involves the ongoing iterative advancement expected to be on display at the RSNA meeting.

CAD offerings build on clinical performance

Cedara Software is developing a computer-assisted detection system for breast ultrasound. The software, announced as a work-in-progress at the RSNA meeting, is intended to identify the sonographic characteristics of benign as well as malignant breast masses and to classify the extent of malignancy for specific nodules.

The company is evolving four configurations of the breast CAD system, which require FDA approval before commercialization in the U.S. B-CAD will identify features of user-defined lesions that correspond to the BI-RADS lexicon in ultrasound images. B-CAD Plus, in addition to flagging user-defined breast lesions, will use sonographic characteristics to compute the probability of malignancy. B-CAD Pro will identify breast lesions and their features. B-CAD Live will provide all the features of the other CAD options in real-time. 

The Cedara B-CAD technology was based on the analysis of more than 1400 images and data from patients with healthy breast tissue, as well as cases with known clinical outcomes following the diagnosis of a breast malignancy. 

CAD system identifies abnormalities in breast sonograms

Medison has added digital multibeam processing to its SonoAce 8000 SE mid-level ultrasound system. The new capability, shown at the RSNA meeting, increases frame rates during both color and 2D imaging and supports harmonic imaging to enhance clarity and spatial resolution. Features include multibeam color and Doppler imaging and pulsed-wave spectral Doppler. 

The enhanced version of the SonoAce 8000 offers software-based 3D image rendering, called Freehand 3D. The company's SonoView Lite image management software and DICOM compatibility are built into SonoAce 8000SE. The company has expanded on-board data storage with an 80-gigabyte hard drive and optional CD/RW drive.

The new system release is being positioned as an entry-level color system for small private cardiology, general imaging, and ob/gyn offices and clinics. The company is planning commercial release for radiology in early 2005, with a cardiology upgrade planned for the second half of 2005. 

Medison has added digital multibeam processing to its SonoAce 8000 SE mid-level ultrasound system. The new capability, shown at the RSNA meeting, increases frame rates during both color and 2D imaging and supports harmonic imaging to enhance clarity and spatial resolution. Features include multibeam color and Doppler imaging and pulsed-wave spectral Doppler.

Digital multibeam processing augments mid-level ultrasound system

The tug of war for cardiovascular imaging has not been kind to radiology. Although imaging of the heart and vessels originally resided in radiology, it soon became dominated by the echocardiogram and gamma camera and gradually migrated to cardiology. If cardiovascular imaging occurred in radiology at all, it was restricted to the nuclear medicine section.

As a result, hands-on exposure to CT and MR studies of cardiac anatomy and pathology fell by the wayside in radiology residencies. Except for those who have been actively involved in cardiac nuclear medicine, most radiologists have not dealt with cardiac anatomy and pathophysiology in any depth since medical school. Although radiologists review cross-sectional cardiac anatomy on CT and MR as part of the cardiothoracic section of radiology boards, a significant knowledge deficit remains in the general radiology community regarding the heart, coronary vasculature, and cardiovascular imaging.

"It's ironic because cardiac imaging is a huge area-half of the disease in the U.S. is cardiac. Yet radiologists have not been educated in cardiac disease to nearly the extent they have about abdominal or brain diseases. Unless they have been doing a lot of nuclear cardiology, they are lacking cardiac imaging experience in their practice," said Dr. E. Kent Yucel, director of cardiovascular imaging at Brigham and Women's Hospital.

But changes in technology, particularly the rise of multidetector CT scanning, are bringing cardiac imaging back to the radiology fold. And they are creating educational challenges for cardiologists as well.

Cardiologists are experienced in analyzing four-chamber views of the beating heart, but they may not immediately know what they're seeing on CT and MR images. Although cardiologists are well schooled in myocardial structure and function and clinical cardiac aberrations, they can't necessarily apply that knowledge to the interpretation of CT and MR scans. Many cardiologists just starting off with cardiac imaging don't fully understand the mechanics of MR or CT or how to run the right protocols for acquiring these images, said Dr. Norbert Wilke, director of cardiovascular MR and CT imaging at the University of Florida, Jacksonville.

Radiologists, on the other hand, know about the underlying MR and CT physics and the procedures for acquiring pictures, and they know how to view scans. But they are less familiar with cardiac disease and its manifestations. Although they can clearly spot an abnormality in the heart or the great vessels and coronary tree, radiologists don't necessarily come to an appropriate interpretation of it or understand how to put it in the context of the patient's clinical problem, Wilke said.

Nor do radiologists have much direct experience in the details of prepping cardiac patients-administering beta blockers to reduce the heart rate or using information from electrocardiograph leads to adjust scan time.

Some large academic medical centers are stepping in to close the cardiac imaging knowledge gap by offering cardiovascular MR and CT training to cardiology fellows and radiology residents and fellows. But smaller centers are not able to do that, and even some of the larger centers don't have the volume to support such education, said Dr. Pamela Woodard, an assistant professor in the chest radiology section at the Mallinckrodt Institute of Radiology, Washington University.

Also, the 16- and 64-slice multidetector CT scanners used for cardiovascular imaging are relatively new. Few imaging centers have broad experience with the machines.

"You have a fair number of people who have already been through radiology training and need to learn about cardiovascular imaging, along with people who come from imaging centers that don't do a lot of cardiovascular imaging. There is a learning curve, at least until radiology departments and cardiology fellowship training programs come up to speed," Woodard said.

The Manhattan Project is an initiative proposed by the Society of Chairmen of Academic Radiology Departments to address the education gap in cardiac imaging. Through the project, 12 academic medical sites across the country are offering week-long teaching workshops to give radiologists and cardiologists at least initial grounding in cardiac imaging through hands-on instruction, actually performing MR and CT scanning on patients with heart and coronary artery maladies.

Typical are the University of Florida's cardiovascular imaging courses. In addition to full-year, three-month, or monthly minifellowships, a three-day course focuses on advanced cardiac MR and CT imaging. This course includes perfusion and function protocols for rest-stress perfusion imaging and dobutamine cine studies for regional and global myocardial functional assessment. All the courses provide a complete array of postprocessing in a dedicated 3D lab with 12 workstations. Hands-on instruction is provided on how to process, display, and report cardiac imaging studies done by technologists and physicians.

Institutions such as the Cleveland Clinic and Stanford University provide full-year fellowships in cardiac MR and CT imaging for radiologists and cardiologists. Sponsored by the RSNA's Research and Education Foundation, the cardiovascular imaging fellowships allow radiologists to work with experienced cardiologists, radiologists, and technologists on state-of-the-art MR and multislice CT scanners in imaging centers that conduct thousands of cardiac exams a year. The Cleveland Clinic, for example, performs cardiac MR examinations on nearly 2000 patients and MSCT studies on more than 6600 patients per year. Stanford conducts about 15 CT studies a day.

As week-long educational efforts, Manhattan Project courses provide self-improvement opportunities for radiologists to begin building their knowledge of cardiac imaging, said Dr. Dieter Enzmann, chair of radiology at the University of California, Los Angeles.

But their short-term nature means these crash courses must leave out many intricacies of cardiac imaging. Participants get a sort of Evelyn Woods introduction to the subject and need more longitudinal follow-up to gain proficiency, said Dr. Richard D. White, head of cardiovascular imaging in the radiology department at the Cleveland Clinic.

"If somebody comes in with no prior knowledge of cardiac imaging and takes a five-day course or even a 10-day course, it doesn't matter. This person will just get an overview and get oriented. Let's add on a three-month course-he or she will be in a much better position to do at least 60% or 70% of the work. Is this person fully competent for complicated cases? Probably not. The individual would still need additional follow-up and practical training to solve all the problems associated with cardiac imaging," Wilke said.

For the most part, cardiologists and radiologists receive the same training. According to Wilke, 80% of the instruction at the University of Florida overlaps.

At the Cleveland Clinic, fellows in the cardiovascular imaging teaching program sit in on the same didactic lectures radiology residents receive on MR and CT. They participate in major general cardiovascular conferences, such as a local board review course, or major conferences on clinical issues, including atrial fibrillation and congestive heart failure. But the bulk of the training involves working with patients alongside skilled imaging teams and conducting related imaging-based research.

At Stanford, fellows participate in assessment of congenital heart disease, valvular disease, atherosclerosis of the coronary arteries, cardiomyopathies, cardiac masses, and pericardial disease. The fellows rotate though Stanford's 3D laboratory and work with five dedicated 3D imaging technologists who process more than 500 clinical cases a month. They learn how to operate 3D visualization and quantitative analysis hardware and software systems.

Fellowships in cardiac imaging are immensely popular. White fields calls almost every day from someone who wants to join the cardiovascular imaging program, but slots are limited by the lack of funding. In the first year of a cardiovascular tomography fellowship, the Cleveland Clinic was able to teach only three fellows: two cardiologists and a radiologist. Two fellows, both radiologists, are currently being trained.

Interest in cardiac imaging training is growing now that technological advancements have made MR and CT much more useful clinically. But cardiac imaging is no different from other areas of imaging in its educational demands.

"To be an expert in cardiac or cardiovascular imaging, the radiologists who perform the studies need to expand their clinical knowledge. But that's true for other radiology subspecialists, whether they are in pediatric, neurological, abdominal, chest, or musculoskeletal imaging," Enzmann said.

Cardiac imaging is a developing and unfolding field. The imaging protocols that are best suited for capturing a dynamic aspect of cardiac function or evaluating myocardial damage or blood vessel patency are evolving. Although CT imaging tends to be more straightforward than MRI, questions persist about the best way to get optimal studies.

"Neither CT nor MRI is fully push-button. Cardiac imaging is not simply looking at a recipe or finding that if you follow a recipe closely, things will work out," Enzmann said.

Radiologists need to become acquainted, in particular, with techniques that will assure high-quality images, such as decreasing the patient's heart rate with intravenous or oral beta blockers.

Optimization of CT and MR imaging nevertheless tends to be relatively straightforward. More difficult is getting radiologists comfortable with viewing the moving heart, Yucel said. Because radiologists are used to viewing still images of anatomy, making a diagnosis from cine imaging of the heart in motion as well as its different segments takes some getting used to. It requires spending considerable time in an echocardiology lab looking at dozens of pictures of the moving heart to absorb all the issues and learn to recognize motion abnormalities.

Radiologists are accustomed to generating images in slices, but to make a diagnosis of coronary artery disease, they must obtain cross-sectional views along the course of arteries following circuitous paths that vary in every patient. Radiologists can't ask a technologist to generate 100 slices so they can go through them sequentially. They have to interact with a 3D data set on a workstation and view the coronary anatomy after multiplanar reconstruction to look for soft plaque and calcium.

CT scanners also spit out images at different parts of the cardiac cycle. An examination of the heart may consist of 10 sets of 200 pictures, each obtained at a different point in the cycle. While one of those sets of images may show all parts of the coronary arteries in a patient with a slow heart rate, viewing several large data sets on a 3D workstation may be necessary to clear all parts of the coronary artery in patients with irregular or more rapid heart rates, Yucel said.

To help radiologists learn to manipulate such large, multitemporal imaging databases, Yucel's cardiac imaging training program provides access to independent as well as vendor-specific workstation software. Radiologists can become familiar with software processing that matches their equipment at home or learn how to adapt processing to different computer software packages.

Understanding the roles of cardiac imaging modalities is a focus at the Cleveland Clinic's cardiac imaging fellowship program. Noninvasive coronary imaging is already a major application for cardiac CT in clinical practice, White said.

One benefit offered by cardiac CT is a strong negative predictive value, obviating catheterization for patients with no significant coronary artery disease who present with atypical chest pain and/or equivocal stress test results. Cardiac CT can also find early markers of atherosclerosis, such as soft plaque and remodeling, that would not be appreciated if the patient went directly to catheterization. And it can detect a high likelihood of significant stenotic disease that may not have been revealed by other modalities.

This growing niche for cardiac CT does not take business away from the cardiac cath lab. It enhances catheterization activity by confining the laboratory to diagnostic studies that precede interventions such as stenting or full diagnostic examinations necessary for later bypass grafting, White said.

Other roles for cardiac CT involve the evaluation of the venous anatomy in patients with a history of chronic atrial fibrillation who may be considered for a pulmonary vein isolation procedure, postprocedure assessment of venous stenoses, pre- and postoperative assessments of the thoracic aorta in patients undergoing stent-graft therapy, and examination of adult patients with congenital heart disease, pericardial problems, cardiac masses, or scar formation who cannot be studied by MR because of the presence of a defibrillator or pacemaker.

The Cleveland Clinic program stresses the importance of radiologists collaborating with other specialists in clinical teams. White points out that radiologists cannot afford to follow the more traditional route for radiology, which is to sit in their own departments and read scans. Radiologists engaged in cardiac imaging need to adopt the mindset of the clinician, rather than that of the imager, by interfacing with the clinical services when they are caring for patients and being aware of the referring physician's interest.

"The cardiologist doesn't want to be interrupted 30 times a day about a CT for pulmonary vein stenosis for the treatment of atrial fibrillation. Radiologists need to concentrate on the fundamental clinical issue that is being addressed and be ready to review the case when the referring physician comes down to radiology. And if there is something of great clinical concern, then radiologists need to seek out physicians and bring to their attention any bothersome sets of findings. So it's not just getting in the habit but getting the religion that being part of the clinical momentum is a key aspect of the radiologist's job," White said.

If radiologists want to be involved in cardiac imaging, they will have to become more in tune with the clinical implications of their work than they have in the past, he said. They have to be concerned less about doing the imaging and more about what imaging means.

"Cardiac imaging is the area that comes to the forefront right now because the pressures for clinical collaboration there are so vivid. But this is directly extendable to other areas of imaging, such as neurologic imaging, and I wouldn't be surprised if oncologic imaging is going to be feeling the same pressure," White said. "Cardiac imaging is taking the lead and showing that it's time for radiology to change its behavior pattern."

Programs fill education gap in cardiac imaging

Pharmaceutical research has brought few biologic drugs to market, despite scores of basic scientific discoveries keying off the sequencing of the human genome. In the U.S., submissions of new molecular entities to the FDA fell from 30 to 20 between 2000 and 2002. Meanwhile, the cost of launching a new drug has increased 55% in the last five years.

The drug development process continues to stall, largely because assessments of the safety and efficacy of experimental therapeutic compounds are antiquated. Makers of oncologic agents rely on guidelines for measuring changes in tumor size, such as the Response Evaluation Criteria in Solid Tumors (RECIST) and World Health Organization standards for calculating tumor dimensions on CT or MR scans.

But these guidelines don't tell pharmaceutical researchers how to evaluate pockets, cystic structures, or tumors that necrose or cavitate in the lung, said Dr. Theresa McShane, worldwide clinical technical director for Pfizer Global Research and Development. She spoke at the 2004 meeting of the International Society for Magnetic Resonance in Medicine in Kyoto, Japan.

Nor can conventional measurements of tumor shrinkage provide a timely determination of treatment response from new classes of therapeutic agents, such as antiangiogenic compounds, which attack not tumor tissue but the blood vessels that feed the tumor, said Dr. Greg Sorensen, director of the Center for Biomarker Imaging at Massachusetts General Hospital.

To avoid financially disastrous drug failures in phase III trials, pharmaceutical companies are optimizing their choices of clinical imaging techniques to facilitate go/no-go decisions in phase I and phase II studies. Even a 10% improvement in predicting failures before entering into phase III trials could save an estimated $100 million in development costs per drug, according to the FDA.

Oncology drug manufacturers are beginning to apply structural, molecular, and functional imaging modalities as quantitative research tools in clinical trials to examine apoptosis, tumor vasculature, and viable tumor volumes and to quantify a drug's mechanism of action, said Dr. Patricia Cole of Novartis Pharmaceutical, at the ISMRM meeting.

In trials of its oncologic agents, Novartis is imaging properties of tumor tissue known as hallmarks of cancer: proangiogenesis, hypermetabolism, hyperproliferation, invasion, and metastases.

"If there is a major alteration in a cancer hallmark, that drug is more likely to show efficacy in larger numbers of patients in later phase trials," Cole said. "So imaging that shows a major perturbation in a cancer hallmark can be very useful in decision making."

In testing an antiangiogenic drug directed at a tyrosine kinase inhibitor at the vascular endothelial growth factor (VEGF) receptor, Novartis employed dynamic contrast-enhanced (DCE) MR to examine colorectal cancer metastases in the liver. After one dose of the drug, imaging revealed a marked reduction in the baseline contrast enhancement of a tumor mass located at the dome of the liver in one patient. On study day two, imaging across all patients showed a change in the baseline transfer coefficient of Ktrans, which was consistent with a dose response. At two months, patients had stable disease.

These findings allowed Novartis to make two important decisions. First was to take the drug forward. If trials had indicated only that patients had stable disease with no evidence of a pharmacodynamic effect, the company would have given up on the drug. Second was to pick an optimal biologic dose. The dose of 1000 mg not only produced significant changes on MR imaging; it also correlated with the actual plasma concentration of the drug.

Four clinical oncology drug development programs at Pfizer have successfully employed DCE MR to studies of drugs in the pipeline. In a dose-escalation phase I trial of Pfizer's VEGF tyrosine kinase inhibitor targeted at tumor vasculature, company investigators performed DCE at baseline, on day two, at the end of the first 28-day treatment cycle, and at week eight. They calculated changes in Ktrans as well as the area under the curve over time (see related story on page 6).

Imaging is becoming such an important aspect of drug development that the FDA, Harvard University, and the Massachusetts Institute of Technology are reviewing its use in pharmaceutical decision making to find specific technologies that could improve or accelerate the process, said Dr. Jerry Collins, director of the Laboratory of Clinical Pharmacology in the FDA's Center for Drug Evaluation and Research.

Researchers from Mass General and the FDA are analyzing data from the published literature and new drug applications to determine how imaging can be used to choose the optimal dose for a drug, based on its presumed mechanism of action. This work is leading to examples that clinical trial sites may follow when conducting phase III investigations.

"There has been an explosion in the number of sites that have molecular imaging technologies, so the technical ability to do these kinds of studies is taking off," Collins said. "We can sit and wait until that translates into NDAs or get ready now and help shape the field."