Pictures of the Future Fall 2008

The Future of Medical Imaging

Medical imaging is helping to detect diseases earlier than ever before. On tap are infrared-based systems that pinpoint abnormal tissues and cells, blood tests that detect traces of cancer proteins, research that is zeroing in on imaging the first signs of Alzheimer’s, and strategies for accelerating the process of discovering new drugs.

Bathed in the glow of a parabolic surgical reflector, the small tumor and its associated lymph nodes were invisible. A previous magnetic resonance imaging (MRI) scan indicated the exact location of the cancer. The question now was which—if any—among the 30 lymph nodes in the patient’s breast, was most likely to have been seeded with cells from the patient’s primary tumor? The surgeon was about to find out.

Pressing firmly against the patient’s skin until she felt the tumor, the surgeon injected a fluorescent liquid into the growth. As she did so, a new imaging system known as Fluorescence-Assisted Resection and Exploration (FLARE) designed by John V. Frangioni, M.D. PhD that employs unique medical image fusion and visualization software from Siemens, combined a visible light image of the area of interest with an image of the invisible infrared light reflected from the fluorescent substance. The image appeared in real time on a nearby color monitor. The result was a clearly visible concentration of brightness at the tumor as well as a river of light moving from it beneath the skin. Following the tumor’s otherwise unknowable drainage path, the river flowed into a nearby lymph node. If any cancer cells had managed to migrate from the tumor, that’s where they would be.

Within minutes the surgeon had resected both the tumor and the glowing lymph node, knowing that in all likelihood she had detected and removed any stray cancer cells at the earliest possible moment and had spared the patient what would otherwise have been major surgery.

The operation, which was first performed at Boston’s Beth Israel Deaconess Medical Center (BIDMC) in July, 2008, marked the clinical introduction of the FLARE imaging system created by Frangioni.  Designed to detect the location of so-called sentinel lymph nodes—those into which a tumor drains directly and that are therefore most likely to harbor any cancer cells that may have migrated from the tumor—Fluorescence-Assisted Resection and Exploration is one of the first applications of the emerging field of optical imaging, a domain that promises to shed new light on shallow-tissue pathologies.

"FLARE in particular, and optical imaging in general hold the potential of detecting the locations of cancer cells before they have a chance of colonizing distant tissues," says Frangioni (see interview), an associate professor of medicine at BIDMC and inventor of the FLARE imaging system, which uses a software platform developed by Siemens Corporate Research (SCR). He adds that, because FLARE makes the exact location of many shallow-tissue pathologies immediately evident, it could significantly cut the time for such procedures. "This could add up to significant savings, because operating room time typically costs $40 to $50 per minute," he says.

A Spectrum of Applications. Frangioni and SCR associates Fred Azar, PhD and Ali Khamene, PhD, point out that fluorescent detection of sentinel lymph nodes is just the beginning. "In addition to detecting anatomical connections between tumors and lymph nodes as is currently the case with FLARE, optical systems can detect a spectrum of physiological processes that are indicative of cancer," says Khamene. "These include changes in oxygen saturation, hemoglobin and water concentrations in tissues, long before any anatomical or structural changes are visible to a surgeon’s eye."

Such a powerful tool could have far-reaching consequences. By providing feedback within just a few hours regarding a tumor’s response to a new medication, for instance, in vivo optical imaging could inexpensively accelerate and personalize drug testing as well as patient treatment.

With this in mind, SCR researchers are working with the Beckman Laser Institute at the University of California at Irvine to develop a novel software imaging platform for a handheld laser and broadband diffuse optical spectroscopy probe that would work in much the same way as does an ultrasound transducer—but with light instead of sound. The device "would be applied directly to the surface of the breast, where it will emit light at many wavelengths and analyze diffuse reflectance, enabling the quantification of intrinsic physiological properties such as oxy- and deoxy-hemoglobin concentration, or water and lipid content," says Azar. "This is a promising application for the future because it will provide immediate feedback as to a tumor’s response to chemotherapy."

Bye-Bye Biopsies? For all its promise, optical imaging is not a panacea. Because it is based on light, it is currently limited to applications involving shallow tissues or open incisions. "In other words, the challenge," says Frangioni, "is to miniaturize optical imaging tools to the point that they can be used in the endoscopic environment—the large intestine and bronchi, for instance—while at the same time developing IR-visible chemicals that make tumor cells visible even if they look normal to the naked eye."

Although these goals will take at least five years to achieve, Frangioni has already come up with the next step in optical imaging: the development of a completely novel "automated microscope"—a device that combines imaging in the near infrared (NIR) part of the spectrum with the gold standard in pathology: H&E (hematoxylin and eosin) staining. The staining procedure is used by histopathology labs in determining a wide range of tissue characteristics. The resulting information supports optimized post-operative patient management.

Frangioni’s automated microscope can take a specimen such as a resected prostate, dice it into 1,000 five-micron-thick slices, stain the slices with an NIR-fluorescent marker substance used in staging cancer cells, and scan the slides with a resolution of one micron at several wavelengths. The microscope’s image processing functions will be driven by a unique software platform now under development by SCR’s Azar and Khamene. "The platform will set the stage for joint analysis of in vitro—microscopic—and in vivo—macroscopic—data," says Azar, who explains that it will assemble the entire dataset derived from a pathology sample and will spatially fuse it with a corresponding 3D MR image in order to pinpoint the locations of different cancer cell types—something that’s never been done before. "The result," says Azar, "will be an unparalleled risk analysis tool supporting evidence-based decisions regarding follow-up treatments."

What’s more, thanks to advanced learning-based algorithms from Siemens, which will be tested at BIDMC, the new microscope "could automate the detection and classification of tumor cells with a very high level of confidence," says Azar. "This would push early detection of cancer cells to new heights because no pathology department has the resources to comprehensively analyze biopsy specimens and investigate everything that looks the least bit suspicious," he points out.

In addition, as more and more MR and histopathology datasets are fused, researchers are laying the groundwork for discovering hidden pattern and tissue characterization information in MR images—information that, according to Frangioni and Azar, may soon make it possible to detect early-stage cancers with MR alone. Could this lead to virtual pathology? "Absolutely," says Azar.

Small Worlds—Early Warnings. While magnetic resonance imaging based on histopathology information holds the prospect of earlier, much less expensive and far more exact detection of cancers than is currently possible, a number of other strategies are now being developed that promise to push detection of several major diseases to an even earlier stage.

In Boston, at Massachusetts General Hospital’s Center for Molecular Imaging Research (CMIR), for instance, support from Siemens is helping Prof. Ralph Weissleder, who directs the Center (see Pictures of the Future, Spring, 2007, interview Weissleder), to develop what he describes as the ultimate early-disease-detection system: a laboratory the size of a pinhead that could be injected into the body and would continuously search for biomarkers indicative of cancer, atherosclerosis, and other major diseases. "We are developing implantable biosensors capable of detecting a spectrum of biomarkers," says Weissleder. "It will be just like having an alarm system inside your body."

The long-range idea is that when such a system detects a disease biomarker, it will also identify it and wirelessly notify the user. "A blood test might then confirm the presence of the biomarker, and the resulting information would make it possible to sequence the disease protein, copy it, produce monoclonal antibodies from it and use these to carry specialized substances to a nascent tumor first to visualize it, and then to destroy it," says Christian P. Schultz, PhD, from Siemens’ Molecular Imaging Division, who heads the company’s Strategic Alliance with the CMIR.

These specialized substances could be molecules such as iron oxide nanoparticles that can be imaged with MR, radioactive tracer materials such as fluorine-18 that can be imaged with PET, or fluorescent molecules that can be visualized with optical imaging. "So there are three routes to discovery and they are totally complementary," says Weissleder. "I believe that each of these technologies will find its niche for specific applications, and that a given patient may wind up being tested with all three."

A Blood Test for Cancer. But before circulating biomarkers can be detected by a futuristic implantable lab or a conventional blood test, these harbingers of disease must be discovered and identified. That’s where Oncogene Science of Cambridge, Massachusetts, a part of Siemens Healthcare Diagnostics, comes in.

The lab is renowned for its invention of a patented, FDA-approved blood test—the only blood test that can monitor the level of the HER-2/neu protein. Elevated levels of the protein are associated with aggressive breast cancer and may be a key player in the ability of tumor cells to divide and multiply. Originally limited to the detection of metastatic breast cancer, the HER-2/neu test has been shown to be clinically useful in the detection of primary cancer—an earlier stage of tumor development during which much less of the protein is released into the blood.

Because of HER-2/neu’s growing importance as a biomarker for breast cancer—as well as indications that it may have diagnostic value with regard to several other major cancer types—pharmaceutical companies are becoming increasingly interested in working with Siemens to test their new therapeutic drugs. "The reason for their enthusiasm is clear," says Head of Oncogene Science, Walter P. Carney, PhD. "Instead of having to test thousands of patients to form a purely statistical basis for the efficacy of a breast cancer medication, we have proven with two HER-2/neu-specific cancer drugs that if a woman’s HER-2/neu level is reduced by over 20 % in response to a medication, she will respond well to that specific drug. This is important because it means that our method allows pharmaceutical companies to get their drugs approved in less time and with fewer patients."

What’s more, by developing tests that can predict whether a medication will work for a specific patient, Siemens could tap into a potentially huge new market known as companion diagnostics. "This is an area in which diagnostics and therapeutics are converging," explains Lance Ladic, PhD, strategic development manager for healthcare at Siemens Corporate Research. "The idea is that before a drug can be prescribed, a diagnostic test should first be conducted to determine if the patient will respond to it. This is a potential growth area for Siemens and represents an entry point into the lucrative pharmaceutical market, a business with global revenues currently exceeding $600 billion."

Companion diagnostics are also good news for patients, as they enhance selection of optimal therapies and reduce adverse drug reactions. Adverse events cause more than two million hospitalizations and 100,000 deaths annually in the U.S. alone, costing the healthcare system approximately $100 billion, according to the Journal of the American Medical Association.

Precisely Targeted Medications. Not only is Siemens advancing the early detection of diseases through biomarker discovery and companion diagnostics, it is also applying much of the resulting knowledge to the development of targeted treatments. "What we want to do is to integrate circulating biomarkers with imaging biomarkers and targeted therapeutic drugs," says Carney. "For example, we expect the HER-2neu test to evolve from a monitoring test for early detection of cancer recurrence—its current status—to an imaging agent that will make it possible to see it in vivo, and finally to medications that can use an HER-2neu-derived monoclonal antibody to target the tumor cells that produced the HER-2neu in the first place. We believe Siemens can be a leader in delivering personalized medicine by providing the right diagnostics to the right patients at the right time."

That vision may sound like it’s still miles away. But in Los Angeles, at Siemens’ Biomarker Research lab, researchers have developed a PET-based imaging biomarker that binds to particularly malignant tumors. By doing so, it makes the tumors visible—and reveals them for what they are much sooner than would otherwise be possible. The compound is based on the same biomarker that is used in a blood test developed by Walter Carney’s group in Cambridge. "Together, the blood test and related imaging biomarker will be important because they target a protein that is generic for all aggressive cancers," explains Hartmuth Kolb, PhD, vice president of Siemens Molecular Imaging Biomarker Research.

Kolb’s lab, which specializes in developing compounds for PET-based imaging, is also exploring strategies for detecting cell proliferation—a characteristic of all malignant cancers. One such strategy is to attach the positron-emitting radioisotope fluorine-18 (F18) to thymidine (a building block of DNA), generating F18-fluorothymidine (FLT), inject it, and image its localization with PET. Since tumor cells synthesize much more DNA than normal cells because of their fast replication, radioactively-labeled FLT is preferentially absorbed by these cells, which results in a bright spot on a PET scan that indicates the location of a malignant tumor. "This work is designed to help diagnose cancers early as well as to monitor response to therapy," says Kolb. The FLT compound, which is now on its way to a Phase II trial, has already drawn the interest of several pharmaceutical companies, which are using it to evaluate new cancer medications.

In addition to its work in early detection and imaging of cancers, Siemens’ Biomarker Research lab is a leader in research related to Alzheimer’s disease. "We are working on the development of PET biomarkers for the detection of Alzheimer’s disease," says Kolb. "This opens the door to early detection of the disease, and allows objective comparisons of new therapeutic substances."

Kolb is convinced that "in about ten or fifteen years" medical science will have advanced to the position where it can keep Alzheimer’s at bay. "Drugs for controlling it will be developed, and their effectiveness for individual patients will be ascertained with PET imaging. But we will need to develop an early warning test to detect it before it causes irreversible brain damage," he says.

Turbocharging Imaging Research. As the detection of diseases has moved from organs to cells, our ability to spot pathologies at an earlier stage has progressed steadily. However, research has been hampered by the heterogeneity of the imaging environment where, particularly when it comes to the crucial transition from animal models to human testing, formats, data sizes and software are often worlds apart. With this in mind, researchers led by Gianluca Paladini at SCR’s Imaging Architectures Program—working under a contract from the U.S. National Institutes of Health’s Cancer Biomedical Informatics Grid program—developed XIP (Extensible Imaging Platform), an open platform that, for the first time, offers a standardized basis for analyzing images from any source, be it cellular, histopathological, preclinical, or traditional radiology.

XIP is made possible by a new, modular plug-in architecture for imaging software that allows thousands of modules to be pieced together to form applications or even entire imaging workflows. This framework supports breakthrough multi-resolution imaging technology developed and patented by Siemens, which makes it possible to integrate and correlate microscopic (for example, in vitro) and macroscopic (for example, in vivo) imaging data.

XIP, which is compatible with the international DICOM (Digital Imaging and Communication in Medicine) standard, will allow researchers and clinicians "to effortlessly integrate images from the full spectrum of modalities and developmental environments," explains Paladini.

"XIP is a game changer," adds Frank Sauer, head of SCR’s Imaging and Visualization Department. "By creating an open, standardized environment for imaging, it will make it possible for a pharmaceutical company to, for instance, offer a new experimental drug with an associated plug-in software model. The software will analyze all the images produced while testing the drug, be they from PET, optical, MR, or other sources. This will lead to much faster analysis of the resulting information, and significantly accelerated drug development times, thus supporting earlier detection of diseases."
                                                                                                              Arthur F. Pease