Cancer diagnosis has long relied on imaging anatomy to identify tumors. With breast cancer, clinicians typically look for structural anomalies on x-ray images. By the time a tumor is large enough to be evident, however, the cancer might have progressed to a relatively late stage. Recent research has therefore focused on functional imaging of possibly cancerous tissue, to identify tumors before they are visible on x-rays. Formation of tumors involves changes in vasculature, which induce changes in blood content and oxygenation that investigators and clinicians can measure with near-infrared spectroscopy. Thus, they can identify tumors even in the earliest stages of formation, which may save lives.The spatial resolution of near-infrared spectroscopy is inherently limited, however, rendering precise localization of the tumors difficult. For this reason, researchers have developed multimodal approaches that combine near-infrared spectroscopy with other imaging modalities, such as MRI, x-ray projection imaging and computed tomography. These techniques offer higher spatial resolution than near-infrared spectroscopy and can provide structural guidance for that method’s functional imaging.Combining modalitiesIn the April 15 issue of Optics Letters, investigators with Dartmouth College in Hanover and Dartmouth Medical School in Lebanon, both in New Hampshire, reported a study in which they demonstrated MRI-guided near-infrared spectroscopy of breast tumors. The team has been working to combine the two modalities for the past several years, said Colin M. Carpenter, the first author of the study. “The key to this study was finally getting all the pieces together — the algorithms, the MR, the optics and the facilities — so that we could operate in the MR and co-register the two.”The researchers obtained the near-infrared spectroscopy signals simultaneously with dynamic contrast-enhanced T1 MRI. The optical information was taken by covering the surface of the breast with 16 fiber optic bundles 13 m long made by CeramOptec of East Longmeadow, Mass., transmitting light from each of the fibers individually and detecting the light at all other fibers in parallel. This arrangement gave them 16 sources and 15 detectors, and thus 240 data points. They repeated the measurements at six discrete wavelengths, from 660 to 850 nm. The illumination sources were laser diodes from various companies; the detectors were Hamamatsu photomultiplier tubes.Researchers have combined MR and optical techniques to provide structural guidance for functional imaging of breast cancer by near-infrared spectroscopy. Shown here are MR images of the region of interest in a breast (top row) as well as reconstructed near-infrared spectroscopy images, from left to right, of hemoglobin (units of micromolar), oxygen saturation (percent), water fraction (units of percent), effective scatterer particle size (units of nm) and scatterer particle number density.After acquisition of the images, the scientists segmented the MR image by region into adipose, fibroglandular and tumor tissue and input these into the algorithm used to reconstruct the near-infrared spectroscopic image of blood oxygenation. They then superimposed the latter onto the MR image.The team encountered a number of challenges during the study. One of the primary difficulties, Carpenter said, was how best to combine the MR breast coil with the optical interface. “The optical imaging worked, but the space constraints inside the MR breast coil were challenging.” The researchers tested a variety of designs before they settled on the one used in the present study. They are still working to improve it, however.The researchers achieved simultaneous acquisition of MR and near-infrared spectroscopy signals by incorporating a fiber optic near-infrared interface into an MR breast coil. The space constraints inside the coil made this challenging, and the team is still working toimprove the design.The other main challenge was how best to co-register the MR and optical data. “We have looked at many different ways to combine these data,” Carpenter added, “and are still looking for the optimal solution. In the end, we feel that the solution will depend on whether the technology is used for screening — where the goal is detecting lesions (reducing false-negatives) — or for diagnosis — where the goal would be to analyze whether a tumor is malignant or not (reducing false-positives).”Clinical applicationIn the Optics Letters study, the researchers demonstrated the efficacy of the multimodal approach, generally, by imaging a 29-year-old woman with an infiltrating ductal carcinoma 2 cm in diameter. The results helped to localize the tumor and showed that it had increased total hemoglobin, decreased oxygen saturation and increased water content — all of these changes are consistent with the higher vascularity and poor drainage of the tumor tissue. The study also revealed that the tumor had increased scatter density and effective scatterer size, thus suggesting that scattering as well can contribute to cancer screening and diagnosis.The investigators continue to develop the combined optical-MR technique with an eye toward clinical application. One of the next goals, Carpenter said, is to develop a patient interface that will accommodate a larger number of women. “We hope that, if we can make this technique more accessible, we will more easily build a large study population so we can look at the bottom-line effectiveness of the device.” Beyond this, the team is working to develop software that is user-friendly for clinicians.Finally, they are exploring other ways to combine the anatomical and functional abilities of MR with the optical sensitivity of near-infrared spectroscopy. For example, broadband and fluorescence MR-guided optical imaging could yield more information with which to compare normal and diseased breast tissue and provide additional means to identify tumors.