The availability of fluorescence standards is necessary in the development of systems and methods for fluorescence imaging. In this study, two approaches for developing diffuse fluorescence materials to be used as standards or phantoms in diffuse fluorescent tomography applications were investigated. Specifically, silicone rubber and polyester casting resin were used as base materials, and silicone pigments or TiO2/ India Ink were added respectively to vary the optical properties. Characterization of the optical properties achieved was performed using time-resolved methods. Subsequently, different near-infrared fluorochromes were examined for imparting controlled and stable fluorescence properties. It was determined that hydrophobic fluorophores (IR 676 and IR 780 Iodide) suspended in dichloromethane and hydrophilic fluorophores (Cy5.5 and AF 750) suspended in methanol produced diffusive silicone and resin fluorescent materials, respectively. However only the hydrophobic fluorophores embedded within silicone resulted in the construction of a material with the characteristics of a standard, i.e. stability of fluorescence intensity with time and a linear dependence of normalized fluorescence intensity to fluorophore concentration.
© 2007 Optical Society of America
Fluorescence imaging has emerged as an important modality for in-vivo observations of molecular function from whole animals and entire tissues. Most notably, fluorescence imaging of small animals is showing great promise as an investigational tool for biomedical research and drug discovery [1–3]. Due to the relatively low absorption of light by tissue in the near-infrared, light can penetrate for several millimeters to centimeters in tissues and, if of appropriate wavelength, excite various fluorochromes present. It is then possible to utilize tomographic methods to reconstruct the bio-distribution of appropriately engineered fluorescent molecules with the ability to tag different cellular targets in-vivo [4–8].
The major complications that arise in in-vivo fluorescence imaging are associated with the high tissue scattering of photon propagating in tissues and the spatial variation of tissue optical properties due to the different structures and organs. These factors result in a strong non-linear dependence of the detected photon intensity on the depth and tissue optical heterogeneity and a reduction on the resolution achieved compared to imaging of transparent media. However, by modeling photon propagation as a diffusive process, a good approximation to the general photon propagation characteristics can be achieved. Combined with tomographic methods utilizing measurements at different illumination projections, fluorescence imaging can then become quantitative and improve the resolution achieved. In addition, methods for modeling photon propagation in tissues assuming solutions of the radiative transport equation have also been suggested [9, 10]. Collectively, these tomographic approaches offer three-dimensional imaging capability and further aim to account for the effects of optical tissue heterogeneity on signal strength and overall image quality.
Besides the diversity seen in methodologies for modeling photon propagation in tissues, there is also significant variation in inversion strategies used for image reconstruction including several different solvers of the forward problem constructed, such as analytical or numerical solutions and iterative non-linear solutions. Similar variation exists in the implementation of experimental tomographic systems developed for in-vivo fluorescence tomography. Systems vary, for example, in their geometrical arrangement by placing sources and detectors on opposite sides of tissue in a slab geometry, or at the same side of the tissue implementing the so-called “reflectance” geometry. Placing sources and detectors around the tissue of investigation, thus offering 360° coverage, is also common. Likewise, some implementations utilize light of constant intensity, whereas others use ultra-short photon pulses at the picosecond range or light of modulated intensity at the 100 MHz – 1 GHz range .
The large implementation possibilities of systems and methods employed in fluorescence tomography of tissues direct the need to develop fluorescence materials that can be used to produce controlled phantoms or standards. Standards are widely used in many fields of scientific instrumentation  and can be employed in system calibration, or the comparison of different system or methodology implementations. For fluorescence imaging of tissues, standards and corresponding imaging phantoms, with accurately determined or controlled diffusive characteristics in addition to stable fluorescent characteristics, become important. Furthermore, the availability of fluorescent materials with different optical properties is critical in fluorescence imaging of tissues, as it is necessary to examine the relative performance of imaging methods as a function of spatially changing absorption and scattering properties in retrieving true fluorochrome concentrations. This is in contrast to more traditional biomedical instrumentation that operates on linear responses, for example in the measurement of fluorochromes distributed in clear materials in photo-spectrometers, common plate readers or other types of monolayer cellular assays [10, 11].
To date, no diffusive fluorescence materials with the properties of standards are available. Instead, a solution consisting of Intralipid and India Ink and a fluorescent organic dye is typically used as a fluorescent diffusive medium. However, optical experiments using these solutions can produce variable optical properties because Intralipid is perishable and its optical properties are unstable as a function of time and temperature. In addition, phantom construction requires special glass or plastic chambers that create unnecessary and potentially optically problematic physical barriers between the diffusive medium and air and complicate the orientation and shape versatility in phantom construction.
To improve on the stability and flexibility in constructing phantoms of any shape and heterogeneity it is important to research solid materials that can be used to impart controlled absorption, scattering and fluorescent properties. Relevant literature has described the use of polyester epoxy and polyurethane resins as appropriate materials for tissue-like absorption and scattering phantoms in the near-infrared. Firbank and Delpy described the use of polyester phantoms as an appropriate solid phantom for absorption and scattering spectroscopy or imaging applications . Polyurethane materials have also recently been characterized and proposed as a suitable reference standard due to its long term stability and negligible effects on the absorption properties through the curing process .
Herein we consider the development of absorptive and scattering materials that can further impart controlled fluorescence characteristics to be employed as fluorescence standards and in the construction of solid fluorescent phantoms. This study describes two methods for producing solid diffusive fluorescent materials and discusses our findings and their implications in the development of diffuse fluorescence imaging.
2 Materials and methods
2.1 Base material
In this research solid diffusive materials with easily manipulated optical properties were of interest. Specifically, two different base materials were used: i) Ecoflex™ 0030 Supersoft silicone rubber Part A and B (Smooth-On, Inc, Easton, PA), and ii) polyester casting resin (Castin’Craft, ETI, Fields Landing, CA). To vary the optical properties of the silicone rubber, white and black silicone pigments (Smooth-On, Inc, Easton, PA) were added. To vary the optical properties of the polyester casting resin, TiO2 (Sigma-Aldrich Chemical Company, Inc, Milwaukee, WI) and India Ink were added. The aim was to devise a recipe which would create a close match between the optical properties of these materials and the optical properties of mice in the near infrared. Recently characterized in-vivo optical properties of nude mice at 732nm, using time resolved methods , have shown that the absorption coefficient μa typically varies in the range 0.15 -0.60 cm-1, and the reduced scattering coefficient μs’ varies in the range 10-35 cm-1, offering significant variability for different body areas. Accordingly, several samples of each material were made with differing quantities of absorbing and scattering particles in order to determine the necessary amounts needed to produce the desired optical coefficients, although only a subset is described herein for brevity.
Silicone materials were prepared by adding different amounts of black and white silicone pigments (as described in results) to 80.0 grams of Part A. The mixture was then stirred and degassed until reaching homogeneity. Part B was then added in a 1:1 ratio of Part A, mixed fully, degassed and poured into a mold. Resin materials were prepared by first mixing together scattering (TiO2) and absorption particles (24.6mg India Ink per mL of styrene) with methanol (1-2% total volume of resin). This mixture was then sonicated for 10-15 min and added to the polyester resin, stirred and degassed. Polyester catalyst (Castin’ Craft, ETI, Fields Landing, CA) was then mixed with the solution according to directions and poured into a mold. Samples prepared for fluorescence measurements remained in the mold throughout the study since data was acquired from the samples during curing (i.e. fluorescence stability over time) and thus the samples required a rigid structure. Samples prepared for optical characterization were removed from the mold so that they could be appropriately placed into the time-resolved system for measurements of the corresponding absorption and scattering coefficients.
2.2 Fluorochromes and solvents
In considering fluorochromes appropriate for constructing solid phantoms, we selected fluorescent dyes that yielded seamless mixing properties with the corresponding base material. In particular, the hydrophobic fluorophores IR-676 Iodide (excitation peak 676nm, emission peak 700nm) and IR-780 Iodide (excitation peak 780nm, emission peak 799nm; Sigma-Aldrich Chemical Company, Inc, Milwaukee, WI) were dissolved in dichloromethane for the production of homogeneous fluorescent silicone standards. Conversely, the hydrophilic fluorophores Cy5.5 (Amersham Biosciences, Piscataway, NJ excitation peak 675 nm, emission peak 694 nm) and AF 750 (Invitrogen, Carlsbad, California, excitation peak 749, emission peak 775) were dissolved in methanol to produce homogeneous fluorescent resin standards. Mixing homogeneity was ensured by observing that photon patterns through the materials matched the theoretically predicted CW photon propagations through a slab. The fluorescent dyes utilized are summarized in Table 1.
2.3 Optical coefficient characterization
Quantification of the optical properties of the different materials produced was performed with a previously reported time-resolved imaging system , displayed in Fig. 1. In short, a pulsed femto-second laser (MaiTai, Spectra-Physics, Mountain View, California) operating at 732 nm was scanned across the sample over a 1cm × 1cm field of view in 2mm steps. The transmitted light from each laser pulse (pulse-width approximately 100 fs) was detected by a high speed CCD camera (LaVision Picostar HR12, Goettingen, Germany) on the other side of the sample, i.e. in trans-illumination mode. The time-dependent images collected at different time gates (200ps gate width, 25ps step) were then analyzed using homemade computer software to obtain time-resolved photon profiles for different radii from the source position . These curves were fitted to the time- and spatially dependent diffusion equation derived for propagation in a homogeneous slab geometry , yielding the absorption and scattering coefficients of the material. The measurements were performed on resin and silicone materials cast in circular laboratory plates with dimensions of 10 cm in diameter and 13 mm in thickness.
2.4 Fluorescence measurements
All fluorescence experiments in this study used a setup based on a previously reported fluorescence imaging system , modified to employ non-contact illumination and detection technology, similar to the system used in §2.3, but operating using lasers in constant wave (CW) mode. This CW fluorescence scanner is capable of epi-illumination, trans-illumination and tomographic imaging. For this study, only the trans-illumination capabilities were used. In these experiments a sample holder substituted the imaging chamber in the original system  and a focused laser beam was scanned on the back of the material using an x-y translated fiber using two linear translation stages. Specifically, light from a 671 nm diode laser and a 750 nm diode laser (B&W Tek, Newark, DE) was routed to a 2×1 optical switch (Dicon FiberOptics, CA) and was focused onto the materials surface using a 100× focusing lens. Samples were securely fixed to the optical table using a clamp apparatus to ensure reproducibly between experiments (i.e. positioning and illumination). Trans-illumination images were captured using a CCD camera (VersArray 1000 Roper Scientific, Trenton, NJ) with the appropriate fluorescence or intrinsic filters. The filters for each combination have been summarized in Table 2. Silicone and/or hydrophobic dye samples were imaged by placing 4mL of the materials in 4mL clear glass threaded vials (Fisher Scientific). Resin and/or hydrophilic dye samples were imaged by placing 2mL of the materials in 3mL semi-micro clear plastic cuvettes (Fisher Scientific). Prior to imaging, the sides of both containers were painted black to minimize lensing effects.
2.5 Auto-fluorescence measurements
The baseline auto-fluorescence of the resin and silicone materials mixed with the corresponding absorbing and scattering particles and their containers, was determined by capturing their fluorescence and intrinsic images without fluorescent dye addition. These measurements were performed immediately after preparation and then 24 hrs later, i.e. after the materials were fully cured. These measurements were done at both the wavelengths of interest (i.e., 672 and 750 nm) and were repeated three times.
2.6 Fluorescence intensity
In order to examine the effects of mixing and curing on the fluorescence intensity for the different materials developed, measurements on free-dye and on cured materials were performed. Fluorescence intensity as a function of fluorochrome concentration was studied for free dyes at 100nM, 400nM, and 1000nM fluorochrome concentrations and at 100nM, 200nM, 300nM, 400nM, 500nM, 750nM, and 1000nM when the dyes were mixed with their corresponding base materials. The measurements were performed immediately after material preparation and also 24 hrs later. These measurements were performed for all combinations of base material and fluorochromes employed.
Free dye was diluted in 1% Intralipid solution and 50ppm India Ink solution with optical properties of μs’ = 12cm-1 and μa = 0.2cm-1. To dilute the hydrophobic dyes (IR-676 and 780 Iodide) into 1% Intralipid, a common surfactant, Trition X-100 (Sigma-Aldrich Chemical Company, Inc, Milwaukee, WI), whose absorption is negligible at the wavelengths of excitation, was added to create a 2.5% detergent Intralipid solution. In this case the hydrophobic dyes were dissolved in dimethyl sulfoxide (DMSO). Correspondingly, for phantom measurements, fluorescent dyes were added to silicone and resin materials of similar optical properties to the Intralipid used (as summarized in the results and Table 3), selected from a larger number of base materials constructed.
2.7 Fluorescence stability over time
The fluorescence stability of the silicone and resin materials was investigated by creating a set of materials with concentrations of 500nM and 1000nM of the corresponding fluorescent dye. These samples were then imaged during the curing/polymerization process, and then periodically thereafter. Specifically, all fluorescence standards were imaged immediately after preparation and then approximately every 30 min for 4-5 hours. All samples were then imaged every 2 days for a total of 2 weeks and approximately every month thereafter for a total of 2 months. When not measured the samples were kept in a dark container and stored in a refrigerator.
2.8 Data collection and processing
Measurements at both the excitation and emission (fluorescence) wavelengths were obtained using the statistics feature within WinView32 (Roper Scientific, Version 220.127.116.11). The pixilated region used to obtain the measurements was the same for all of the samples in the experiments, and was inspected each time to ensure that no pixel saturation was present. The value obtained was the average intensity in counts. Background noise was subtracted from this measurement, and it was then converted to counts/sec. To obtain quantitative measurements, the emitted fluorescence was divided by the incident excitation light. This normalization eliminates possible variations in the fluorescence intensity recorded due to variations of light intensity, which was especially important considering the measurements were performed over a period of two months. The normalization also significantly minimizes the sensitivity of the result of the division to possible inaccuracies in the exact optical properties achieved in each material constructed .
3.1 Optical coefficients
Figures 2 and 3 illustrate the dependence of the optical properties of the base materials on the amount of pigment/particle added. Specifically, Fig. 2 depicts the optical properties of the silicone phantoms as a function of the amount of white and black pigment contained in the sample, while Fig. 3 shows the optical properties of the resin phantoms as a function of the amount of India Ink and TiO2 contained in the sample. As observed in Fig. 2, the scattering properties of the silicone material were controlled uniquely by the amount of white pigment added, while the absorption was dependent on both the amount of black and white pigment added (i.e. the white pigment contributed to both the scattering and absorption of the phantom at the wavelengths used). Conversely, Fig. 3 shows that the absorption and scattering properties of the resin materials could be controlled independently by the amount of India Ink and TiO2 added, respectively.
In both cases, a single material mixture was selected for the remaining fluorescence studies. The selection criterion was that the phantoms contain the necessary amount of absorption/scattering particles to yield μa≈ 0.2cm-1 and μs’ ≈ 12cm-1, which is representative of the lower-end optical coefficients found in mice . To obtain these optical coefficients for silicone rubber; 1.3mg of white pigment and 0.13mg of black pigment must be added for every 0.5ml of both parts A and B (i.e. 1 ml total). Furthermore, to obtain approximately the same coefficients for resin, 1.5mg of TiO2 and 0.74mg of India Ink diluted in styrene was added for every 1ml of resin. Table 3 provides a summary of the optical properties for the selected silicone and resin materials and the corresponding materials utilized to impart the optical properties observed.
Figure 4 depicts the auto-fluorescence recorded from the base materials mixed with the absorber/scatterer concentrations described in §3.1 and their corresponding containers. The measurements were obtained before and after curing at both wavelengths of interest. There is a considerable increase in the amount of auto-fluorescence at 750 nm compared to 670 nm. However, the auto-fluorescence observed in either wavelength is at least two orders of magnitude weaker compared to the values acquired at the lowest fluorophore concentrations (i.e. 100nM) used in this study (see results in Fig. 5), and can be practically considered negligible. Additionally, no significant differences in auto-fluorescence were recorded before and after curing.
3.3 Fluorescence strength of free fluorochromes and fluorescent materials
Figure 5 depicts the free fluorescence strengths for all of the fluorochrome concentrations investigated within this study when dissolved in the Intralipid solution described in §2.6. As observed, titrations of free dyes revealed that all fluorochromes studied exhibited a positive linear dependence of normalized fluorescence intensity vs. fluorochrome concentration. Furthermore, all trends fit a linear model well (i.e. high coefficients of determination, R2) given a zero y-intercept, with the minor deviations observed attributable to minor pipetting errors. Referring to Fig. 5, AF 750 [Fig. 5(b)] produced markedly higher normalized fluorescence intensity values when compared to the other fluorochromes [Figs. 5(a), 5(c), 5(d)], whereas Cy5.5 [Fig. 5(a)], IR 676 [Fig. 5(c)], and IR 780 [Fig. 5(d)] yielded more comparable fluorescent signals for a given concentration. No significant differences were observed with repeated titrations or with measurements taken 24 hours after initial mixing.
Correspondingly, Fig. 6 depicts the fluorescence strength measured for the varying concentrations of fluorescent dyes mixed into the base materials. As observed, the recorded normalized fluorescence intensity was proportional to the concentration of the fluorophore in the sample for all dyes. Notable differences were observed between the hydrophilic [Fig. 6(a) and 6(b)] and hydrophobic dye titrations [Fig. 6(c) and 6(d)]. Linear regression of the Cy5.5 titration data [Fig. 6(a)] and AF 750 data [Fig. 6(b)] yielded a good linear fit as expected. However, compared to the free dye measurements, there was significant reduction in the signal from the AF 750 dye, whereas no significant signal drop was observed for the Cy5.5 fluorochrome. As discussed in §3.4 the signal intensities observed did not remain steady over time; instead the titration data obtained after 24 hours showed that Cy5.5 lost 70-80% of the original signal and the AF 750 signal was almost completely degraded.
Linear fitting the data obtained from the titrations of IR 676 [Fig. 6(c)] and IR 780 [Fig. 6(d)] produced regressions that did not pass through the origin. As shown in §3.2 this behavior cannot be attributed to auto-fluorescence of the material but possibly some interaction of the dye with the base material. However, the fluorescence produced in these titration experiments was nevertheless linear. Comparing the titrations between free dyes and base materials, there was a reduction in the measured signal for both IR 676 and IR 780 when the dyes interacted with the base materials. Therefore, the fluorescence dilution in base material needs to be higher than that of the free-dye to result in the same fluorescence strength. However, the values obtained before curing and after curing remained constant for IR 676 in silicone, but a reduction in signal after curing was observed for IR 780 (discussed further in next section). Figures 6(c) and 6(d) show the values measured after the silicone had completely cured for both dyes.
3.4 Fluorescence stability over time
The stability of fluorescence intensity as a function of time for each of the materials constructed is depicted in Fig. 7. The Cy5.5 dye in resin [Fig. 7(a)] showed that there was a 70-80% reduction in intensity in the first 24 hours, and then a less rapid but fairly steady degradation after curing with about a 1-2% decrease in intensity each day. The resin material using the AF 750 dye demonstrated a significantly faster decay, shown in Fig. 7(b), resulting in complete quenching of fluorescence activity during the curing process. At the end of curing only the material’s auto-fluorescence remained.
Hydrophobic dyes placed within silicone provided results that were more appropriate for the construction of stable fluorescent materials. Referring to Fig. 7(c), which is the long term stability of IR 676 Iodide within silicone, no degradation was observed during or after the curing process. Some 10% changes can be observed possibly also to some measurement error. The IR 780 [Fig. 7(d)] demonstrated a 40-45% strength decay during the curing process but after curing produced a material with constant long term fluorescence intensity.
The large range of implementation possibilities for fluorescence imaging systems and the corresponding methods used for diffuse fluorescence tomography dictate a need for fluorescence standards. In this study, we sought to develop and evaluate diffusive fluorescence materials which could be used for accurate evaluation or calibration of systems and methods used in fluorescence imaging of optically opaque media such as tissue. Specifically, we examined two different preparation methods and independently characterized the optical properties, fluorescent strength and long term stability of silicone and polyester resin based fluorescent materials.
Polyester and silicone materials were selected due to several reported advantages for the construction of solid phantoms. These materials are rigid, non-toxic and have easily manipulated optical properties. Furthermore, they can be cast into arbitrary geometries, and machined or cut to create desired shapes. Both materials allow for the production of more realistic tissue phantoms based on the possibility to make regionally different optical properties in the same phantom. One distinct difference between these materials is in the mechanical properties. The silicone material exhibits elastic properties that are more similar to that of real tissue and in certain applications could be more applicable than polyester resin, which produces a non deformable rigid material.
Characterizing how different amounts of pigment changed the optical properties of the materials revealed that polyester resin was more suitable for the creation of optically specific materials. The addition of India Ink and TiO2 acted independently towards the absorption and scattering properties of the polyester resin, respectively (Fig. 3). Conversely, whereas the black pigment acted solely on the absorption of the silicone material, the white pigment contributed to both the absorption and scattering properties of the silicone material (Fig. 2).
As such, producing silicone materials with a given absorption and scattering coefficient becomes a coupled two-parameter approach compared to the more simple recipes derived for the polyester resin.
Furthermore, a non-linear chemical/physical interaction appeared to take place between the fluorescent analyte and the silicone base material. Titrations with the fluorophore and silicone could accurately be fitted with linear regressions, which however did not pass through the axes origin [Figs. 6(c) and 6(d)]. This finding was highly counterintuitive especially based on the free dye titrations and auto-fluorescence measurements which showed that negligible signal should be expected at the absence of fluorochrome in the silicone material. Nevertheless, this phenomenon was confirmed by four repeated trials, and was consistent throughout the study.
In observing the long term stability data recorded from the silicone based fluorescent materials, it was found that the IR-676 Iodide was stable when added in the silicone, however the IR-780 Iodide had a reduction in signal intensity during the polymerization process and only produced stable intensity readings after the completion of the curing process [Fig. 7(c) and (d)]. Conversely, the polyester resin was found to be ill-suited for producing fluorescent standards or for use in fluorescent phantoms. Referring to the titration data of Cy5.5 and AF 750 in resin [Figs. 6(a) and 6(b) respectively], linear measurements were obtained when the resin material was uncured. However, when the same measurements were taken post-curing, virtually complete fluorescence signal loss was observed that rendered the material incompatible with the fluorophores studied and therefore unsuitable for fluorescence investigations.
The results shown herein employed normalization of fluorescence measurements to the corresponding measurements at the excitation wavelength. This approach minimizes the fluorescence strength dependence on the laser intensity fluctuations that may be observed between measurements or possible coupling issues that may differ from measurement to measurement. Additionally, the ratio measurement is less sensitive to the exact optical properties of the material, compared to the uncorrected fluorescence signal and therefore yields more accurate measurements of fluorescence strength. It is known however that some dependence of the ratio on the optical properties exist , in particular ~10% overestimation of the ratio is expected for ~2x increase in background attenuation. The use of the ratio was nevertheless essential throughout the study, especially since measurements over long periods of time and variations in the laboratory systems employed could be expected. The utility of using the ratio is shown in Fig. 8. Figure 8(a) plots the raw counts for the measurements in Fig. 6(c). The fluorescent measurement obtained for the 200nM concentration (oval region) initially appears to be an outlier; however, it can be observed that a corresponding increase was observed in this measurement at the excitation wavelength. With appropriate normalization of the fluorescence measurement to the excitation light measurement; the intensity and concentration became linearly related in all cases as shown in Fig. 8(b).
In summary, the results showcased that IR dyes, placed in silicone, produced fluorescent materials that would be suitable as diffusive fluorescent reference standards or for creating stable fluorescent phantoms. The resulting fluorescent material was found to be both stable and reproducible and can be made to match the reduced scattering coefficient and the absorption coefficient of mouse tissues in the near-infrared. In contrast, resin based materials were found significantly inferior in terms of fluorophore stability in the base material, and as such, inappropriate for producing diffusive fluorescent materials.
The authors would like to acknowledge Antoine Soubret, Nikolaos Deliolanis and Yuan Hushan for their valuable and always available insight, the staff of the Harvard-MIT Division of Health Sciences and Technology Biomedical Optics Summer Institute and its director Dr. Charles Lin. This work was supported by the US Army and Materiel command grant W81XWH-04-01-239 and the National Institutes of Health grants R43-ES012360 and R01-EB00750.
References and links
2. V. Ntziachristos, J. Ripoll, L. H. V. Wang, and R. Weissleder, “Looking and listening to light: the evolution of whole-body photonic imaging,” Nat. Biotechnol. 23, 313–320 (2005). [CrossRef]
3. M. Gurfinkel, S. Ke, X. X. Wen, C. Li, and E. M. Sevick-Muraca, “Near-infrared fluorescence optical imaging and tomography,” Disease Markers 19, 107–121 (2003).
4. M. J. Eppstein, D. J. Hawrysz, A. Godavarty, and E. M. Sevick-Muraca, –Three-dimensional, Bayesian image reconstruction from sparse and noisy data sets: Near-infrared fluorescence tomography,– Proc. Natl Acad. Sci. USA 99, 9619–9624 (2002). [CrossRef]
6. H. B. Jiang, S. Ramesh, and M. Bartlett, “Combined optical and fluorescence imaging for breast cancer detection and diagnosis,” Crit. Rev. Biomed. Eng. 28, 371–375 (2000). [PubMed]
7. S. V. Patwardhan, S. R. Bloch, S. Achilefu, and J. P. Culver, “Time-dependent whole-body fluorescence tomography of probe bio-distributions in mice,” Opt. Express 13, 2564–2577 (2005). [CrossRef]
10. A. D. Klose, V. Ntziachristos, and A. H. Hielscher, “The inverse source problem based on the radiative transfer equation in optical molecular imaging,” J. Comput. Phys. 202, 323–345 (2005). [CrossRef]
11. U. Resch-Genger, K. Hoffmann, W. Nietfeld, A. Engel, J. Neukammer, R. Nitschke, B. Ebert, and R. Macdonald, “How to improve quality assurance in Fluorometry: fluorescence-inherent sources of error and suited fluorescence standards,” J. Fluoresc. 15, 337–362 (2005). [CrossRef]
12. M. Firbank and D. Delpy, “A design for a stable and reproducible phantom for use in near-infrared imaging and spectroscopy,” Phys. Med. Biol. 38, 847–853 (1993). [CrossRef]
14. M. Niedre, G. Turner, and V. Ntziachristos, “Time-resolved imaging of optical coefficients through murine chest cavities,” J. Biomed. Opt. in press (2006). [CrossRef]
15. M. S. Patterson, B. Chance, and B. C. Wilson, “Time resolved reflectance and transmittance for the noninvasive measurement of tissue optical-properties,” Appl. Opt. 28, 2331–2336 (1989). [CrossRef]
16. E. Graves, J. Ripoll, R. Weissleder, and V. Ntziachristos, “A sub-millimeter resolution fluorescence molecular imaging system for small animal imaging,” Med. Phys. 30, 901–911 (2003). [CrossRef]
17. V. Ntziachristos and R. Weissleder, “Experimental three-dimensional fluorescence reconstruction of diffuse media using a normalized born approximation,” Opt. Lett. 26, 893–895 (2001). [CrossRef]
18. A. Soubret, J. Ripoll, and V. Ntziachristos, “Accuracy of fluorescent tomography in the presence of heterogeneities: Study of the normalized Born ratio.” IEEE Trans. Med. Imaging 24, 1377–1386 (2005). [CrossRef]