Optical Imaging Workshop from Bench to Bedside at the NIH

September 7-8, 2000
National Institutes of Health
Bethesda, Maryland

Summary

The following is a summary of speaker presentations and panel discussions that occurred during the 11/2-day conference from September 7-8, 2000. This summary attempts to capture salient points noted by speakers and to identify common themes and issues raised by participants in the panel discussions.

Thursday, September 7

Welcome
Deborah Hirtz, NINDS
Ed Staub, NIH Clinical Center, NCI)

Dr. Staub emphasized clinicians' need for real-time information on the structure and function of human systems, as well as effects of disease; optical imaging holds promise in meeting these needs. The challenge of the workshop was to identify when optical imaging will be ready for the patient; what needs to be done to make optical imaging more useful to the patient; and, finally, how these goals will be achieved.

Dr. Hirtz then welcomed participants, thanking workshop sponsors (NICHD, NCI, NHLBI, NEI, NINDS, ORS, CIT, ORWH, OSA, Hamamatsu, ISS, OSA, and Zeiss), and began with a brief overview of the 1999 in-vivo optical imaging workshop. The 1999 workshop sought to present the latest developments in optical imaging methods, address theoretical and technological limits of in-vivo optical imaging methods, identify technological developments with the greatest potential for clinical and research applications, and distinguish the potential for optical imaging to offer something above current technologies. New, versatile, efficient, and non-invasive techniques discussed at the conference included multiphoton microscopy, confocal microscopy, optical coherence tomography, contrast agents, and multispectral imaging. Additionally, participants at the 1999 conference discussed technical and commercial factors leading to success for new modalities, in addition to diffuse imaging and techniques to overcome scattering effects. Finally, Dr. Hirtz listed suggestions of the 1999 panel, which included recommendations that technologies developed for imaging should be easy to use, inexpensive relative to other technologies and targeted to the appropriate market. At the same time, these technologies should have sufficient sensitivity and specificity for clinical usefulness while addressing a clinical need.

Biomedical Optics as the Translational Research Ideal
Bruce Tromberg, University of California, Irvine

This session was designed to inspire participants about imaging, as well as define translational research. Dr. Tromberg first presented a broad overview biomedical optics, defining it as the ability to develop technologies to view gene expression in single cells, use light to view remitted light signals that travel through several-centimeters-thick tissue, do high-resolution imaging of endogenous structures without exogenous probes, and add new contrast elements. Dr. Tromberg emphasized that biomedical optics is both a diagnostic and a therapeutic field; diagnostics require light absorption or scattering, while therapeutics require only light absorption in the tissue. Biomedical optics are effective because of fundamental light/tissue interactions, which are due, perhaps, to the unique spatial, spectral, and temporal properties of lasers (e.g., lasers beams are capable of focusing high over small spaces, and can launch light through relatively thick tissues). Translational applications of optical imaging in the brain are relatively new. One application is in using optical coherence tomography to image the three-dimensional structure of the cortex. In addition to specific applications biomedical optics allow for sophisticated probe designs, by modifying surgical instruments. Barriers to translational research include energy, distance between laboratories, and lack of funding. Teamwork and a process of interlaboratory linkage can help overcome these barriers.

Necessary and Sufficient Conditions to Bring a Technology from Bench to Bedside
Warren Grundfest, UCLA

Dr. Grundfest stated that there are no hard and fast rules by which to transfer technology from bench to bedside, but rather a series of lessons learned from experience. First, there are several technologies that are dramatically changing the face of health care; for instance, advances in imaging are leading to development of minimally invasive surgical and nonsurgical therapies. This, plus a goal of controlling health care costs, has led to a push toward outpatient versus inpatient procedures. Additionally, therapies are moving out of the sophisticated atmosphere of the operating room and intensive care unit to the doctor's office. Also, patients are not always sent to the imaging modality, rather, the imaging is brought to the patient.

Dr. Grundfest also discussed some critical imaging technologies, including functional interventional magnetic resonance imaging; high-resolution, three-dimensional ultrasound; and the development of imaging tools that contain the entire optical imaging system. Technology development requires several parallel activities to succeed: a clear statement of the problem, identification and protection of intellectual property, review of competing technology, understanding the optimal status, and understanding the FDA approval process. In addition, key personnel and skills must be identified, costs must be understood, and financial resources must be identified. Successful technology development requires effective implementation of an organized research and development plan incorporating all of these activities.

Question and Comments

Participants were encouraged to comment on both presentations. Questions focused on how to motivate physicians to adopt new technology and how to finance technological advances. Other questions dealt with spatial and temporal claims, software development, and the importance of collaboration.

Advances in Optical Imaging of the Brain
Daniel Hanley, Johns Hopkins University

To understand brain imaging, it is necessary to understand the major chromophores in the brain: oxyhemoglobin, deoxyhemoglobin, and cytochrome AE3. Because the brain is an oxidative organ, these chromophores can be exploited. Contrast agents are used to transmit time measurements during circulatory paradigms, and can also be used in infrared images. A model that views reversible changes in cytochrome AE3 was used to identify large concentrations of hemoglobin inside the skull. This model showed that there are changes in hemoglobin oxidation, which means that algorithms can then predict brain saturation of hemoglobin and detect hematomas. Fluorescent gene reporters also can be identified through noninvasive techniques. Light scattering may provide unique information about subcellular particles through a system that assesses the transmission of scattered light through the brain slice and uses optical theory to measure particle size. Recommendations for future research include making optical systems available to practitioners and modifying the system based on practitioners' experiences. Proof-of-principle studies should have increased funding, as should clinical trials of spectroscopy and imaging in non-imaging environments.

Optical Methods and the Brain: Potential Role in Scientific and Clinical Neurology
Arno Villringer, Charite, Berlin

Dr. Villringer reviewed the various types of information that can be obtained from use of optical methods. He also discussed obtaining information non-invasively from intact human brains and meeting the needs of brain research, in particular, clinical neurology. Currently, the concentration of certain chromophores can be measured, most importantly, those that have distant absorption patterns in the near-infrared are the ones whose concentration can be determined. Dr. Villringer cautioned that when determining concentrations, it is critical to take into account factors such as scattering events. Dr. Villringer explained that scattering can convey cellular information, for instance, about the electrical state of individual cells. It is possible to visualize the brain non-invasively because in the near-infrared there is an optical window that has very little absorption and can penetrate 10 centimeters of tissue. Experiments have confirmed that it is possible to measure signals non-invasively in human and animal models. Near-infrared spectroscopy has been successful in localizing or classifying epileptic seizures as well as, but no better than standard methods. It is unlikely that optical monitoring for stroke patients would be successful. The likelihood of clinical success is higher for detecting intracranial hemorrhage, especially subdural and epidural hemorrhage, and for monitoring hemoglobin saturation. Application of optical imaging in the clinical setting will depend on whether applications are needed to penetrate far into the brain, or whether visualization of the cortex is sufficient.

Panel Discussion
Beniamino Barbieri, ISS, Inc.; Britton Chance; David Benaron, Stanford; Daniel Hanley; Debra Hirtz; Arno Villringer

Participants discussed potential uses for the imaging technologies presented during the sessions. Some of these applications include monitoring development of cerebral palsy and other applications in children and babies. In addition, participants discussed encouraging information gathering on optical imaging technology. For instance, stroke is a prevalent disorder; yet there is no EKG that allows titration of therapy, as there is for the heart. One panelist emphasized the importance of not relying solely on imaging, but also using other indicators of a patient's condition whenever possible.

Optical Coherence Tomography in Ophthalmology
James Fujimoto, MIT

Dr. Fujimoto discussed his work in optical coherence tomography (OCT) in the context of developing a technology and transferring it into a clinical setting. Dr. Fujimoto helped develop an optical biopsy that examines tissue microstructure in situ and in real time, using interferometry to measure high-speed echo time delays of optics. The goal is to construct a cross-sectional image of the anterior or posterior eye by measuring back reflections along an axis. This approach gives a view of the retinal pathology, at much higher resolution than ultrasound, and is not invasive. In order to have clinical impact, the instruments must be accessible to the clinical community. This technology was transferred to Humphrey Instruments in 1993 and introduced to the marketplace in 1996. OCT has several clinical indications, including staging of macular holes and quantitatively measuring the thickness of the retina and the nerve fiber layer in glaucoma patients. This concept of high-resolution imaging of structure can also be applied to fields other than ophthalmology in cases where standard biopsy is not possible. The advantage of non-invasive imaging is that repeated imaging can be performed over time on the same subject. The challenge of getting the technology to the clinic has been great. Relative to other applications of fiber optics, this application is not profitable, and the path to market is long, mirroring the challenges that the biomedical optics industry faces.

Clinical Confocal Microscopy of the Cornea
Barry Masters, University of Bern

The clinical confocal microscope, co-developed by Dr. Masters has raised the question, "What have we learned that we didn't know before development of the new instrument?" This microscope scans the back focal plane of the objective. Light then comes back through the prism with two confocal slits, is descanned, and comes up to the detector, providing a real-time, real-color image of the eye or brain. A database has been compiled of several thousand patients that have been examined using the microscope. In addition, images from the front to the back of the cornea are videotaped and linked with the patient's record. To answer the question of what has been learned from the microscope, researchers conducted a double blind randomized trial of patients who wore and did not wear contact lenses. Patients who wore contact lenses more than 10 years developed microdots of lipofuscin on every cell in the stroma. This condition degrades visual acuity and contrast. Thus, by using the confocal microscope, researchers discovered a corneal disorder that had not been observed previously. In his concluding remarks, Dr. Masters emphasized that it is important for the next development in clinical confocal microscopy to use non-contact devices.

Adaptive Optics for the Human Eye: Pushing the Limits of Retinal Imaging
Austin Roorda, University of Houston

When imaging the human retina, clinicians are forced to look through its optics, which are clear. However, diffraction and wave front aberrations degrade retinal images. Moreover, aberrations vary from one individual to another, corrections must be made for each eye imaged. Using a deformable mirror to correct, and using a flash lamp to illuminate the retina, an extra-sharp image can be obtained on a CCD camera. Using this imaging technique, researchers can see nearly every cone receptor, as well as where receptors are missing or not functioning, allowing researchers to measure properties of the photoreceptors. The angular tube properties of an array of cones also can be determined by measuring how much light is reflected from the array as a function of the incident light angle. This technique has been applied clinically to one woman who presented with 20/200 vision in the central 5 degrees of both eyes. Using the imaging technique, it was determined that the vision problem was due to missing or non-functioning photoreceptors in that area. Therefore, in addition to being used as a diagnostic tool, laser treatment is also a potential application of this technology.

New Scanning Laser Systems for Diagnosing Eye Disease
Andreas Dreher, Laser Diagnostic Technologies, Inc.

In order to bring a technology from bench to bedside, there are four steps that must be followed. First, a researcher should determine if there is a need that must be met and then discover a solution that meets the need. When the solution to the problem has been determined, a prototype should be developed, including regulatory approval, feedback from scientists and clinicians, accuracy and reproducibility, and comparison to existing methods; then the product is marketed to clinicians. Finally, the developer needs to have the financial justification for reimbursement. As an example, Dr. Dreher described the process of designing a new scanning laser tool to diagnose glaucoma. Approximately half of all people with glaucoma do not know they have the disease even when current diagnostics are used, therefore, it is clear these diagnostics need improvement. Thus, the research team decided to devise an instrument that looks directly at the nerve fiber tissue that is the cause of glaucoma, rather than using more indirect standard diagnostic techniques. This instrument uses a scanning laser system to get real-time data capturing and imaging, which allows reconstruction of a live image through the undilated pupil. Clinical studies show that the instrument is 96 percent sensitive and 93 percent specific. In addition, patients prefer this test because it only takes about 1 second to complete, whereas standard techniques can take nearly an hour. The first bench model was tested in 1989, received FDA approval in 1993, and the first models were delivered to researchers and scientists. Based on their feedback, the device was revised and a clinical model was produced in 1999. The clinical model was introduced in 2000. Dr. Dreher also briefly described a screening test for age-related macular degeneration, which is not as far into the development process as the first device.

Imaging from the Bench to the Retina Specialist and from the Bench to the Family Doctor
Ran Zeimer, Johns Hopkins

Many patients with degenerative eyesight could be better treated if diagnosed earlier. To solve the problem of late diagnosis of diabetic retinopathy, screening should be brought to the patient, rather than waiting for the patient to come to the screening. Dr. Zeimer described a digiscope that images the back of the eye. This instrument is used in the primary care practitioner's office when the patient presents for regular visits. The physician captures data, which are then read by experts; the physician can then refer the patient to an ophthalmologist, if necessary. The digiscope performs comparably to the standard-stereo fundus photography-and early results show that screening rates can be doubled using this tool. To illustrate the process of transferring technology from the bench to the ophthalmologist (rather than the family doctor), Dr. Zeimer described an instrument that measures the thickness of the retina over the posterior port. This instrument produces a map of the thickness of the retina. This map is compared with a "normal" map as well as maps from the patient's previous visits to determine the extent of any deterioration. This technology has been applied in diagnosing macular edema. The technology stands now with the most promising technologies in the field and faces the same problems: small market, acceptance, and reimbursement.

Panel Discussion
Andreas Dreher, James Fujimoto, Barry Masters, Michael Oberdorfer, Austin Roorda, Ran Zeimer

Panelists discussed issues relating to clinical application of new optical technologies. New technologies may have applications in multiple specialties; because of this lack of focus, clinical development can be difficult. To better impact clinical care, development should follow a business model, considering outcomes, cost, and where to implement technology that is being developed. Caution should be taken to ensure a balance between basic and applied science, however, since many agencies choose to fund studies with concrete end products, rather than those with basic science as their goal.

What Is Missing from Current Imaging Modalities in Breast Cancer?
Rosemary Altemus, NCI

New developments in breast cancer therapy allow delivery of well-defined radiation dose with a rapid fall-off so that normal tissue receives very little therapeutic dose. This increased treatment precision leads to increased necessity to define the target area. Current imaging does not provide the needed information on physiologic changes, metabolism, or vascular supply. When defining target treatment areas, clinicians need images that provide biologic and mechanistic data that includes metabolic, biochemical, physiologic, and functional categories. Biological images can yield genotypic and phenotypic information that can aid in genetic and molecular diagnosis, as well as gene therapy. Clinicians are now beginning to get information such as blood flow, hypoxia, and pH from the MRI MRS spectrum of information, which can help lead to diagnosing malignant versus benign tumors. In addition, PET scans are versatile, highly sensitive, and provide very good imaging effects. In one pilot study PET scans were equivalent to sentinel node biopsy in detecting malignant tumors in the breast and axillary lymph nodes. Enhanced imaging can also help in administering gene therapy, for instance, by determining when gene expression is maximal, as well as the best time to initiate treatment. Imaging strategies are needed to be able to view, on a cellular level, all possible biological and functional aspects.

Breast Imaging: Applications and Opportunities
Mitchell D. Schnall, University of Pennsylvania

Current screening tools for breast cancer detect abnormalities in the breast, however, not all of these abnormalities are necessarily cancerous. Once discovered, abnormalities are followed and diagnosed as benign or malignant. Imaging assists in diagnosis by such techniques as guided biopsy. Imaging also assists in staging by determining the size of the lesion, and is also important in follow-up. Methods for breast cancer imaging must be highly sensitive, moderately specific, generalizable, and patient friendly. Current mammographic methods for finding breast cancer are considered to be 85 percent sensitive; PET and MRI methods, however, find almost twice as many cancer as mammography, therefore the sensitivity of mammography is probably much lower than the accepted rate. New imaging modalities will help screening for breast cancer by showing elements that otherwise could not be seen. For instance, optical imaging can determine hemoglobin concentrations, which can help to differentiate benign from malignant tumors. Additionally, optical imaging, such as optical MRI and ultrasound, can help guide biopsies.

Functional Optical Spectroscopy of Human Breast Tissue
Bruce Tromberg, University of California, Irvine

The risk of developing breast cancer increases more than 10-fold between the ages of 30 and 60, indicating that there must be dramatic changes in breast tissue during that period. For instance, ductal structures enlarge and respond to hormone levels, and after menopause, there is a loss of epithelial tissue in the breast. Typically, mammograms do not perform well on younger women, because of their more dense epithelial structures. Mammograms on older women typically show enhanced contrast between tumors and surrounding tissues. Studying the breasts women of all ages can help determine the biologic age of breast tissue, thus gaining insight into issues such as risk, response to therapy, prevention, and adjuvant therapy. Measurements taken using a photon migration system show the entire absorption and scattering spectra; comparing spectra of premenopausal and postmenopausal women, dramatic differences in composition and saturation levels become apparent. Scattering is much greater as a function of wavelength for premenopausal women than for postmenopausal women, which may be due to the greater abundance of epithelial tissue and collagen in premenopausal women contribute to the scattering. In one study of 21 women, some trends are visible. For instance, from the ages of 20 to 30 years, hemoglobin levels sharply increase, peaking at 30 years of age, and then decreasing. The loss of hemoglobin from ages 30 to 60 years correlates with the increased risk of breast cancer in that same age group. It was also determined that lower scatter power is reflective of higher fat content in the tissue. In addition to helping diagnose tumors, optical techniques may be able to help prescribe chemotherapy. For instance, in one patient monitored during chemotherapy, saturation levels changed throughout the chemotherapy. This could be indicative of a revascularation process or hypoxic zones within the tumor. Malignant tumors and benign tumors both have individual signatures using this imaging, but it is too early to be able to determine from this method alone which are malignant and benign. Studying aging and the effects of hormone replacement therapy will also help gain insight on the progression of tumors and the effects of chemotherapy.

Panel Discussion
Rosemary Altemus; Jeremy Hebden, University College, London; Brian Pogue, Dartmouth; Herbert Rinneberg, Technische Bundesanstalt, Berlin; John Schotland; Mitchell D. Schnall; Bruce Tromberg; Martin van der Mark, Philips Research Laboratories, Netherlands

This panel discussion began with individual presentations from four of the panelists, and was followed by a question and answer session.

Dr. Pogue began by discussing how functional optical spectroscopy can be applied in small-scale clinical trials. To determine the diagnostic value of hemoglobin-based breast cancer imaging and examine contrast mechanisms and multispectral functional imaging via the contrast of hemoglobin, Dr. Pogue uses a fiberoptic array that can interrogate a single tomographic slice of the breast. Preliminary results of the small pilot study show that tumors present with increased blood concentration. Multispectral imaging is necessary to determine oxygen saturation, which is one of the most promising parameters that will be examined in the full study.

Next, Dr. Hebden described a neonatal brain imaging system and its possible application to breast imaging. This system illuminates at one of 32 points on the surface of the patient, and can detect a time-resolved signal at 32 other points. The system uses a series of rings for imaging that can fit over almost any breast, and thus avoid compression of the breast. The reconstruction is based on a three-dimensional FEM model. The system uses reference phantom measurements to get a degree of separation between absorption and scatter. The first clinical subject will be tested with the new system soon after Dr. Hebden's return to London.

Dr. Rinneberg continued the discussion by describing his work on a European project for imaging and characterization of breast lesions by pulsed near-infrared light. His team uses multichannel scanning instrumentation and contrast agents for breast imaging. This instrument has shown that tumors have increased total hemoglobin and oxyhemoglobin saturation, but lower oxygen saturation. The project will establish objective criteria for tumor detection by optical means, which requires clarification of the origin of false positive results that have been seen up to this point.

Dr. van der Mark described the latest results of the Philips optical mammography clinical study. This device uses 3 wavelengths of light at 50 milliwatts to image the breast. A trial studied the instrument on a group of women at increased risk for breast cancer. Standard radiology discovered 29 malignancies in the group of 328 women; the new instrument found some, but not all, of the 29 tumors. Dr. van der Mark pointed out that some of the 29 tumors were not seen on mammogram, but with other methods such as ultrasound. The new imaging device was successful in imaging five occult tumors. Tumors did not display very low oxygenation, as would be expected, but they were distinct features in the image.

During the question and answer session, participants and panelists discussed methods of improving imaging, with one participant suggesting that signals can be improved by having patients breathe high oxygen content gas. This will cause normal tissue to become highly oxygenated, thus making it easier to differentiate from abnormal tissue. Panelists also discussed whether optical mammography would ever be used as a screening tool. One panelist stated that the resolution for imaging may never be good enough for use in screening, and another suggested that optical mammography is likely to be used only as an add-on tool. Other participants and panelists still foresee clinical relevance for optical mammography, but raised caution about creating artificially high expectations for the field. Finally, Dr. van der Mark noted that Philips is withdrawing from the optical mammography market and offered to provide participants with the full dataset, to minimize the risk that it will not be analyzed.

Friday, September 8

Real-Time Contrast Enhanced Imaging of Breast and Prostate
David Benaron, Stanford

Dr. Benaron began by stating that the first limitation to optical imaging is that not every physical or pathological state has an inherent optical signal. For instance, many of the properties that are associated with tumors are bulk phenomena that may not apply to very small tumors. In these instances, contrast agents will help when the native signal is weak, absent, or non-specific. Contrast agents can be directed toward specific tissues, including antibodies against service receptors, which are uniquely expressed on tumors. The goal of contrast is to give the physician or surgeon the same advantages in-vivo that the pathologist has ex-vivo. New methods of targeting (e.g., receptor targeting), have helped improve contrast. In addition, antibodies and antibody fragments are getting more and more specific. Specific dyes, including luciferase can be used to label individual cells, thus assisting in monitoring genetic engineering projects. At this time, this treatment cannot detect spontaneous disease, only treated cells that are introduced into the subject. Optical contrast: has several advantages: it is targetable, tolerable, safe, highly sensitive, inexpensive, not radioactive, and has direct placement for many applications in the radiotracer market. Optical imaging allows better real-time detection than both PET and PCR because it is in-vivo and real-time. Some contrast agents that are being developed for optical imaging include a prostate antigen targeted against an extracellular region of the PSMA molecule, a lymphatic marker that biodistributes, and a colon marker.

Early Cancer Angiogenesis Detection by Optical Imaging with GFP
Mark Dewhirst, Duke University

Dr. Dewhirst began by defining green fluorescence protein (GFP) is a way to image early angiogenesis in tumors and monitor gene therapy. Contrast agents, including GFP help identify tumor margins and vessels. This technology helps researchers view the relationship that exists between tumor cells and endothelial cells. Using skin flap window chambers and GFP to monitor activity in-vivo, researchers monitored the effectiveness of an anti-angiogenetic compound. Experiments showed inhibition of vascularly linked disks, and even complete inhibition of tumor growth. In addition, the technology has shown that angiogenesis in metastatic sites seems to start earlier than previously thought. Using GFP, researchers viewed tumor cells migrating toward vessels in the body, when the mutated X-EPO protein was introduced, angiogenesis was halted. In summary, researchers learned that angiogenesis of metastasis occurs in the 100-200-cell stage. GFP also can be used to monitor the oxygenation of tumors and the earliest stages of angiogenesis to see if hypoxia plays a role in early behaviors. GFP can also monitor gene therapy.

Detection of Cervical Precancers: A Multi-Modality Approach
Rebekah Drezek, University of Texas, Austin

The goal of Dr. Drezek's research is to develop minimally invasive, cost effective tools to improve screening and detection of curable precursors to epithelial cancer. The current screening method, Pap smear, has low sensitivity and specificity. Colposcopy, a screening method used after an abnormal Pap smear, also has low specificity. The time between these tests and receipt of results can be weeks, which can be a cause of anxiety for patients. Optical imaging has the potential for real-time automated assessment, which will be available to more remote locations and help avoid losing patients to follow-up. Real-time imaging would also limit the number of biopsies and treatments, have the potential to monitor biochemical and structural progression, and may even be able to identify lesions that are likely to progress. When developing new technologies, five levels of assessment are necessary: biological plausibility, technical feasibility, intermediate outcome, patient outcome, and societal outcome. Fluorescent spectroscopy, where spectra are collected and their differences analyzed, can help to discern information about the physiological state of the tissue. Spectral differences observed may be due to metabolic state, nuclear size, vascularity, epithelial thickness, and architecture. Fluorescent spectroscopy was compared to the current standard of care, and outperformed several other methods. Clinical trials also show that autofluorescence can provide high sensitivity and specificity, but age and menopausal status influence intensity variations that are found in the data. In addition to its high specificity and sensitivity, pain and anxiety associated with spectroscopy are less than what is found with other techniques. In addition, a "see and treat" diagnostic strategy combining colposcopy and spectroscopy produces cost savings over other techniques. Other than fluorescence, optical technologies being explored as methods to automate and improve screening and detection of cervical precancers include reflectants, polarized reflectants, confocal imaging, optical coherent stimography and macroscopy imaging. Dr. Drezek stated that the optimal approach to imaging might be some combination of several of these methods.

Imaging of Dynamic States by Optical Tomography
Randall Barbour, State University of New York

Dr. Barbour began by discussing how vascular reactivity influences the stability of optical measurements. Because of this, dynamic imaging (which images the time variability and the hemoglobin signal) is used. The standard imaging methodology, x-ray angiography, can detail the architecture of the vascular tree, but, unlike dynamic imaging, provides no information on its functionality. Dynamic imaging requires a reconstruction approach, time series analysis methods, and hardware systems that allow for fast parallel data capture. Adaptive finite element methods improve computational efficiency in reconstruction. Time series analysis allows for examination of pixel data according to various measures. Dr. Barbour described an instrument used for breast imaging that mechanically stabilizes the breast and uses various kinds of light sources, individual photo diode channels and a simple CCD camera. The system uses a geometrically adaptive system to minimize motion artifacts and conform the target to a particular simple geometry. Lock-in detection eliminates the influence of ambient light, improves on noise, and can discriminate more than one wavelength of light simultaneously. Duplex ultrasound has been shown to monitor the time variability of some of the major arteries and tissues, and dynamic imaging is able to do the same kinds of measures, except in cross-sectional view. Dynamic imaging provides robust images with intrinsically high contrast, providing detailed investigations of mechanisms involved in tissue vascular coupling for large tissue structures.

Contrast Mechanisms in Optical Coherence Tomography
Joseph Izatt, Case Western Reserve University

Dr. Izatt discussed OCT's combined confocal and coherence ranging capabilities, which can be used to image in both transparent and non-transparent tissues. Gastrointestinal tissue is one of the most translucent and optically accessible tissues, thus it was chosen as the first model for many investigators. Dr. Izatt stated that preliminary imaging with OCT initially took 10-20 seconds, which was too slow for imaging inside the gastrointestinal tract. Thus, a new scanning technology for a very fast kilohertz rate reference term had to be developed. The instrument developed uses circular technology to recover some of the light coming back from the sample. In one examination of a normal human esophagus, the probe imaged more details of the structure than were imaged in vitro. The real-time imaging shows different layers of the structure, glands, and blood vessels. Compared with the gold standard-histology-the probe underestimated the thickness of the mucosa. This was because these experiments did not control for the amount of pressure on the probe. Dr. Izett hopes to use this technology to image surface abnormalities related to early cancer. The high-speed technology developed for the gastrointestinal endoscopy can also be applied to image the anterior segment of the eye. Range-gated spectroscopy can be used with OCT to extract spectral features of the tissue. This technology is developed enough that it is ready to be applied to a number of applications in dermatology, dentistry, and gastroenterology.

Multi-Modal Miniature Microscopes for Detection of Pre-Cancer
Michael Descour, University of Arizona

The first step to moving technology away from the bench, according to Dr. Descour, is removing barriers among technologies, engineering, life sciences, and institutions. The general method for imaging pre-cancers is a pen-sized probe on which is mounted a miniature complete microscope. Design of this instrument is a collaborative effort among several groups. In order to miniaturize a microscope system, the image plane must get very close to the object. Transverse magnification of the microscope objective must also be minimized. When working in collaborations, several people with differing areas of expertise must work in tandem, thus, the fabrication process must be simplified to eliminate the need for a variety of expertise. In optical systems, a set of tolerances must be satisfied for the systems to deliver useful performance. When systems are assembled within these tolerances, no alignment is required, and the need for expertise is lessened. Based on this simplified assembly technique, fluorescence imaging and optical sectioning can now be accomplished at the same time. To do this, a small grading must be created that can be mounted on a wafer and translated sideways. Several images will be taken at different positions of the grading as it scans; the images will then be digitally processed. In integrated design, the process of designing the optics and building the miniature microscope should not stop at just the optics and components that make up the system, but should also include the contrast agents. Dr. Descour recommended that researchers consider ways to include elements that can implement some kind of translation, in addition to the passive elements in a microscope or endoscope.

Panel Discussion
Randall Barbour, David Benaron, Mark Dewhirst, Rebekah Drezek, Jay Knuston, Joseph Izatt, and Michael Descour

Note: Due to technical difficulties, many participants' questions and comments could not be recorded. Panelists' comments were recorded, however, and are reflected below.

Participants discussed the role of funding agencies, including the NIH, in helping to produce multiple copies of new devices so they will be available for clinical trials. Dr. Barbour expressed his support for instituting some type of mechanism that would allow researchers to make new devices available to other researchers, rather than simply focusing on the commercial market. Participants raised questions about possible applications of new devices that were presented during the sessions, as well as clarification on the modes of operation of the devices. Dr. Knutson concluded by noting patients' fears of needles, and suggested referring to some of the field's abilities as "optical acupuncture," since fibers can be very small.

Panel: Research Resources Needed to Promote the Field of Optical Imaging
Lawrence Clarke (NCI), Chair; Leon Esterowitz (NSF), Co-Chair; William Heetderks (NINDS); Michael Marron (NCRR)

The purpose of this panel was to explore current NIH funding mechanisms and address the issue of technology transfer from bench to bedside. Dr. Clarke, of the Biomedical Imaging Program of the National Cancer Institute (NCI), began the discussion. The Program's initiatives are structured in two phases, the first of which established an organization to bring diverse groups of investigators together, and the second of which creates an organizational structure with specialized resources to conduct feasibility testing of new products. Additionally, the Program has issued an RFA to promote interdisciplinary, shared imaging resources and research related to small item imaging. It is important to share software resources, such as normative databases. One research model that optics could follow is the NCI's new model of funding private industry to build tools (e.g., ultrasound devices), but only if industry then licenses the devices to investigators. In addition, when developing new devices, it is important for investigators to identify the unique buyers of, and requirements for, optical systems.

Dr. Heetderks next discussed NIH Bioengineering Consortium (BECON). As a result of BECON, proposals that are design driven, as well as hypothesis driven are accepted. But NIH is still focused on the research side of research and development, if investigators would like to see a shift in that policy, they must send a clear message to the NIH. However, NIH does support training, e.g., with summer internships and the BECON K25 award that supports training in biomedicine for established researchers with technology backgrounds. NIH also distributes bioengineering grants to support research, in addition to supporting the bioengineering research partnership.

Dr. Esterowitz, of the National Science Foundation (NSF), discussed some of the Foundation's main funding mechanisms. These mechanisms include the engineering research and technology centers. Additionally there is the biomedical-biophotonics partnership, together with the NIH and NSF-106, in which all NSF divisions will participate. Another program that is being considered, the Small Group Program, would fund provide funds ranging from perhaps $100,000 to $300,000, as opposed to the several-million-dollar programs normally funded.

Dr. Marron, of the National Center for Research Resources (NCRR), discussed funding mechanisms within NCRR. NCRR funds resource centers and requires collaboration, service, dissemination, and training from each center. The Shared Information Grants Program is an annual program in midlevel or high instrumentation, with a $500,000 upper limit. In addition, with the NSF, NCRR participates in a program to allow investigators to apply simultaneously for NSF and NIH money for instruments with funding needs that exceed $500,000. NCRR also funds research project grants, including the SBIR and STTR programs, which are combined efforts among commercial entities and research institutions. Finally, the NIH participates in the NSF partnership discussed by Dr. Esterowitz.

Questions and Comments

Note: Due to technical difficulties, many participants' questions and comments could not be recorded. Panelists' comments were recorded, however, and are reflected below.

Panelists and participants clarified and furthered discussion on the funding mechanisms at the NIH. When seeking funding, it is important to stay in touch with program personnel in order to maintain an understanding of policies and procedures. While it is possible to obtain funding to produce just a few copies of an instrument in order to allow clinical study, it is not a priority of the NIH. Investigators should reiterate the importance of this process to the NIH in order to help change its priorities.