Multimodal imaging systems advance breast imaging and other applications. The most widely used technology for early detection of breast cancer – x-ray mammography – still suffers several major limitations. It misses roughly 20 percent of breast cancers, for example, and its performance is even worse in younger women with higher-density breast tissue. Also, the technique exhibits a high false positive rate, with 70 to 80 percent of breast biopsies coming back negative. This can add considerably to patients’ financial and mental stress. Optical imaging can help to address these limitations. Using only non-ionizing radiation, it can detect deep-tissue physiological parameters, such as hemoglobin and oxygenation maps, many of which have been found to be relevant to the presence of breast cancer. However, stand-alone optical imaging techniques have low spatial resolution and thus cannot detect small lesions. Over the past decade, researchers at Massachusetts General Hospital and Harvard Medical School have developed and tested a tomographic optical breast imaging system that can be used simultaneously with a digital breast tomosynthesis system, and potentially also with 2-D mammography scanners, to acquire co-registered structural and functional images. At the same time, the researchers have made progress in developing joint image reconstruction algorithms that use the x-ray anatomical information to enhance optical image quality. DBT = digital breast tomosynthesis; RF = radio frequency; DAQ = data acquisition. Courtesy of Qianqian Fang, Massachusetts General Hospital. For this reason, researchers have been working to combine optical imaging with other imaging modalities. By performing optical imaging together with x-ray techniques, for example, users can get the best of both worlds: the high image contrast of the former and the high spatial resolution of the latter. This integrated approach is called multimodal imaging, and it could lead to important advances in breast imaging and other applications. “By reading x-ray and optical breast images side by side or overlaying them via advanced data fusion techniques, radiologists can easily extrapolate findings from one modality to the other,” said Qianqian Fang, an assistant professor of radiology at Massachusetts General Hospital and Harvard Medical School in Boston who has been developing a combined optical/digital breast tomosynthesis (DBT) technique. DBT is a relatively new method in which an x-ray tube captures multiple images around the breast and uses them to create a 3-D reconstruction. (A quick note of disclosure: Fang works with David Boas, the author’s brother, at Massachusetts General Hospital.) Researchers at the University of California, Irvine, are developing a multimodal system that combines optical imaging – in both absorption and fluorescence modes – with MRI, to improve the specificity of the latter. In fluorescence mode, the system can provide two different images: concentration and lifetime. Lifetime depends on environmentally induced physiological changes such as local pH, blood supply and temperature, which have been shown to vary with the tumor growth and metastasis. Shown here are fluorescence concentration and lifetime images with no a priori (left), with diffuse optical tomography (DOT) functional imaging only (center) and with DOT functional and MRI structural imaging (right). The latter combination gives the best quantitative accuracy. Courtesy of Gultekin Gulsen, University of California, Irvine. Importantly, Fang continued, the structural and functional information can be used together to detect and diagnose cancers. This works almost like a traffic overlay: The x-ray scan provides a detailed image of the structures of the breast – the road map – on which is superimposed information about tissue functional properties in the breast. This information can be, for example, variations in hemoglobin concentrations and metabolic demand reflecting the formation and growth of tumors. Using the approach, clinicians may be able to find more cancers earlier – which can help to save more lives. And it’s not just x-ray tomosynthesis. Researchers are exploring the potential of multimodal imaging in the clinic with optical imaging and no fewer than four other modalities: MRI, PET and ultrasound as well as the x-ray technique. Because of its high sensitivity to tumors, dynamic contrast-enhanced (DCE) MRI is used for detection and diagnosis of tumors in patients at high risk for breast cancer. DCE-MRI acquires a series of images before and after administration of a contrast agent and measures the contrast enhancement kinetics from the tumor. The technique can return a high rate of false positives, though, as the low-molecular-weight extracellular agents used can diffuse freely through tumor blood vessels, leading to an overlap in enhancement kinetics between malignant and benign tumors. Thus, as with conventional mammography, DCE-MRI can result in anxiety for patients subjected to unnecessary biopsy or overtreatment. Using x-ray anatomical information as structural priors has already contributed to improved optical image resolution in the combined optical/x-ray tomosynthesis system. Shown here are total hemoglobin, oxygen saturation and scattering coefficient maps, as well as the original tomosynthesis image, for a breast with a 2.5-cm malignant tumor. New developments in optical techniques – including high-density probes and fast camera-based imaging methods – are expected to deliver even higher image quality. The researchers plan to incorporate these emerging techniques into the system to make it faster, more flexible and more compact. Courtesy of Qianqian Fang. Groups at Dartmouth College and the University of California, Irvine, are working to improve the specificity of DCE-MRI. They are using optical imaging to quantify the optical properties of hemoglobin, oxygen saturation and more to better distinguish between malignant and benign tumors. “The idea is to use the structural a priori info provided by the anatomic modality to guide and constrain the image reconstruction algorithm of functional imaging modality to improve its quantitative accuracy,” said Gultekin Gulsen, an associate professor of radiology and physics at the University of California, Irvine. Optical imaging is desirable as the functional imaging modality because it uses safe, non-ionizing radiation, because it is relatively inexpensive and, most importantly, he said, because of the shelf life of the optical contrast agents – especially as compared with the radioactive agents used with PET, for example. Gulsen and colleagues at the University of California, Irvine, recently published the results of feasibility studies of their MR-FDOT system using small animals (FDOT = fluorescent diffuse optical tomography). They have nearly completed development of a clinical breast imaging interface (shown here) and are planning to begin clinical studies in a couple of months using the nonspecific fluorescent contrast agent indocyanine green (ICG). “Since we can get an image every 16 seconds, we hope to get the enhancement kinetics of the ICG that may differentiate benign and malignant lesions,” Gulsen said. Courtesy of Gultekin Gulsen. Developing multimodal systems will always present unique challenges. Fang notes the need to create an optical design that enables seamless integration with an x-ray mammography system. But integrating optical imaging with MRI has added complexity: namely, the high magnetic field associated with MRI and the tight space within which everything must fit. The high magnetic field means you can’t use metal or other magnetic materials in the scanners. Optical fibers up to 10 m are a common sight in multimodal imaging studies, as the optical imaging instrumentation must be kept in a separate room. The fiber optic interface also must be nonmagnetic. This can be particularly challenging, Gulsen said, because the interface has movable parts and should be adjustable for each patient. “The other challenge is space,” said Michael A. Mastanduno, a graduate student in the Optics in Medicine lab at Dartmouth, which is developing the multimodal breast imaging system. “The way the MR scanner is designed, there’s not a whole lot of space to fit the equipment around the person in the bore of the scanner.” The Dartmouth group is optimizing the performance of the system by collecting data from healthy volunteers. Shown here are combined optical-MRI images of total hemoglobin and other parameters. Courtesy of Michael A. Mastanduno. The Dartmouth group addresses the space issue using long optical fibers and custom designs to integrate them into the MR breast coil. These can be moved to target a variety of tumor locations, depending upon the patient, Mastanduno said. He noted, however, that an ideal commercial system would include an array of magnetically compatible optical detectors housed in a breast biopsy plate, thus eliminating the need for large fibers while providing better detector coverage and adjustability. “That’s our next step,” he said. Multimodal in the clinic? With these various approaches to multimodal breast imaging, which is most likely to succeed in the clinic? No single strategy is viewed as a blanket solution, said Xavier Intes, an associate professor of biomedical engineering at Rensselaer Polytechnic Institute in Troy, N.Y. Intes worked with multimodal breast imaging in the early 2000s as a postdoctoral fellow in the laboratory of the late Britton Chance, renowned professor of biophysics, physical chemistry and radiologic physics at the University of Pennsylvania; recently Intes published a review of diffuse optical tomography integrated with other, clinically established methods. Rather, he said, the potential of each approach is considered in light of how it can contribute to a specific part of the patient management program. Combining optical imaging with MRI can help to advance imaging for breast cancer by overlaying functional information about tumor growth onto an anatomical map provided by MRI. Shown here is a system developed by researchers at Dartmouth College using near-infrared spectroscopy and MRI. Note the 12-m fiber optic cable leading to the optical imaging system outside the MRI scanner room. DAQ = data acquisition; RF = radio frequency. Courtesy of Michael A. Mastanduno, Dartmouth College. Fusions with x-ray can advance screening, for example. MRI and PET come later in the patient care cycle: either in diagnosis, in establishing the site of a tumor prior to surgery, or in evaluating a patient’s response to chemotherapy. Each strategy has its strengths, and each will have a role to play in the clinic. Even having identified where a multimodal breast imaging system will have the most impact, though, clinical translation could prove a particular challenge. There are the usual hurdles, of course, not the least of which is convincing technologically conservative clinicians to incorporate a new tool into their work flow. But developers must also question what kind of market exists for the systems. “When we were first thinking about optical imaging, we were thinking about a technique that cost very little but was widely disseminated,” Intes said. But now, combining the technique with other, considerably more expensive modalities such as MRI, “the market share is going to be completely different.” Researchers are confident that the benefits of multimodal imaging – the improved specificity due to the availability of anatomical information, for example – will ultimately outweigh any concerns clinicians may have about it. Intes points to a 2008 Physics in Medicine and Biology paper that shows shipments of PET-only scanners declining to zero between 2002 and 2007, with the overall market for PET or PET/CT scanners remaining fairly constant. This underscores a recognition of the benefits of the latter, multimodal systems, he said. Extending multimodal’s applications Multimodal imaging is enabling advances in areas beyond the breast. At École Polytechnique de Montreal, Frédéric Lesage and colleagues are integrating ultrasound and fluorescence imaging to study atherosclerosis and cancer in mouse models, and extending preclinical developments to learn about atherosclerotic plaque in larger species using intravascular techniques. Researchers have combined ultrasound and fluorescence imaging to study atherosclerosis and cancer. Shown here are (left) a combined ultrasound-fluorescence image of the heart in a mouse atherosclerotic model and (right) an in vivo intravascular ultrasound-fluorescence image of atherosclerotic plaques in a rabbit model. Courtesy of Frédéric Lesage, École Polytechnique Montreal. As with multimodal imaging in the breast, the anatomical information provided by ultrasound helps to improve the accuracy of fluorescence imaging. And of course ultrasound is a mature technique, already commonly used in cardiovascular disease and oncology. The information it provides on its own can serve as a complement to fluorescence imaging, Lesage said. For applications to small-animal imaging, the researchers have combined fluorescence and ultrasound imaging in a single compact and relatively inexpensive preclinical system with a single-element transducer for ultrasound imaging and a laser source and an electron-multiplying CCD camera for fluorescence imaging. Using this system, they have demonstrated quantification capabilities “on par with or better than more expensive commercial systems.” The investigators continue to develop the combined system. Incorporating 2-D transducer arrays and camera detection will greatly facilitate further research. At the same time, work to broaden this application to larger species has led to additional studies combining fluorescence, photoacoustic and ultrasound imaging. “With multimodal intravascular imaging, similar studies are under way in rabbits to test the different markers and their co-localization with various constituents of the plaque,” Lesage said. “The system has already been tested on over 20 rabbits, but the sensitivity of the fluorescence and photoacoustic components still needs to be increased.”