We describe a fluorescence lifetime imaging technique employing the collimation detection capabilities of an angular filter array (AFA). The AFA accepts minimally scattered photons emitted from fluorophores up to 2 mm deep within turbid media. The technique, referred to as Angular Domain Fluorescence Lifetime Imaging (ADFLI), is described and its performance evaluated in comparison to a conventional (lens and pinhole) system. Results from a tissue-mimicking phantom demonstrated that ADFLI provides better spatial resolution and image contrast for fluorescent probes at greater depths compared to a lens and pinhole system.
© 2010 OSA
The recent advancement of disease-targeted fluorescent markers has stimulated concurrent developments in non-invasive optical molecular imaging techniques. These advancements have made optical molecular imaging an effective alternative to positron emission tomography and magnetic resonance imaging, predominantly in drug delivery research. In particular, time-resolved fluorescence imaging offers a rich array of information, namely fluorescence localization and lifetime that can be used to interrogate local molecular and physiological processes in tissue, providing complex diagnostic and prognostic information [1,2]. The combination of this technique with molecularly-targeted contrast agents having physiologically-sensitive lifetimes has obvious applications for early-stage cancer diagnostics since the distribution of fluorescence could be used to locate tumors and the lifetime of the fluorescence could be used to investigate tumor microenvironment (notable examples include pH and temperature) [3–6]. In addition, fluorescence lifetime imaging (FLI) can be used to distinguish fluorescent probe distributions more effectively than intensity mapping since lifetime can be used to separate out the auto-fluorescence signal . Fluorescent lifetime imaging can be used to spatially resolve combinations of different fluorescent probes with single excitation wavelengths but different lifetimes [7,8]. Recent progress has been made on the application of near-infrared FLI in a high resolution, wide field-of-view imaging platform that enables acquisition of quantitative optical property maps with simultaneous reconstruction of lifetime . Applications of FLI to cell imaging by microscopy have been numerous and highly successful . The use of two-photon microscopy provides the benefit of better localization of fluorescent targets in samples at deeper depths with reduced auto-fluorescence [10,11]. For optically-thick, highly-turbid samples such as tissue, several approaches to FLI have been developed. One approach is to solve the inverse problem using diffuse photon density waves . The fluorescence lifetime can be extracted using time domain (TD) or frequency domain (FD) measurement techniques [13–15]. Compensation for the rapid loss of spatial resolution as a function of depth has been attempted by combining finite-element light propagation models with image reconstruction algorithms [16,17]. However, large depth-dependent discrepancies in signal intensity arising from embedded fluorophores make it difficult to resolve complex distributions containing both shallow and deep fluorescence populations, precipitating the need for multiple projections to be collected . Time domain techniques provide the potential advantage of direct measurement of lifetime by means of multiple exponential fits to fluorescence decays, an aspect well exploited for non-scattering thin tissue samples by fluorescence lifetime imaging microscopy (FLIM). However, in thicker samples, the diffuse propagation of light within the tissue creates asymptotic decay of the TD fluorescence temporal function, which must be corrected with reconstruction methods .
Although much effort has been directed at applications of FLI to microscopy and FLI to imaging macroscopic samples (e.g. tomography), here we present a reflectance-based FLI method optimized for the mesoscopic domain (i.e. beyond the capabilities of FLIM but below the centimeter spatial scale representative of macroscopic imaging). The method is based on angular domain fluorescence imaging (ADFI), which offers a conceptually simpler technique to mitigate spatial resolution degradation with depth and minimize the loss of signal intensity . With ADFI, spatial resolution is improved by restricting the detection of scattered photons arising from the imaging medium with an angular filter array (AFA). The AFA is comprised of a parallel array of high aspect ratio micro-tunnels, micro-machined into a silicon substrate . The objective of the work reported here was to extend the capabilities of ADFI to mesoscopic fluorescent lifetime imaging. The paper builds upon our preliminary work [22–24] and begins with an overview of angular domain fluorescence lifetime imaging (ADFLI) followed by a comparison of the technique to FLI with a Keplerian lens and pinhole. The comparison was achieved by successively imaging tissue-like phantoms containing two fluorescent sources with distinct lifetimes. Experiments were repeated for sources placed at various depths within the phantom.
2. Angular domain fluorescence imaging
In ADFI, an angular filter array is employed to reject unwanted scattered photons and accept the spatially informative quasi-ballistic photons based on acceptance angle. As shown in Fig. 1(a) , the angular filter device differentiates between minimally scattered fluorescent photons from the more plentiful scattered fluorescent photons emitted from fluorophores situated in tissue. The scattered photons emitted from fluorophores have less likelihood of exiting the tissue surface with an angular deviation acceptable to the AFA. Only photons emitted from the fluorophore that have small angular deviation are accepted by the AFA. Hence, these minimally deviated photons can pass through the AFA and be detected by the camera. Since the performance of the AFA is not dependent on coherence, or the wavelength of light, and AFAs can be fabricated several centimeters in width, ADFI can be used to map fluorophores over a large field-of-view with millimeter range spatial resolution at significant depths (~2 mm) into tissue .
The angular filter array consists of a parallel array of square-shaped micro-tunnels (typically 80 microns wide along a 1.5 cm long plate) to obtain a high aspect ratio of approximately 188:1 (see Fig. 1(b)). The bottom component of the AFA is fabricated by etching a silicon substrate and the walls of the bottom component can be patterned with many small features to suppress internal reflections within each micro-tunnel (Fig. 1(c)). A flat silicon wafer is used as the top component to enclose the micro-tunnels to form the AFA. Since the micro-tunnels are square in geometry, there exists an angular acceptance angle variation from 0.3° (wall to wall) to 0.42° (corner to corner). This design is known from previous work to be selective for quasi-ballistic photons and provides at least 200 µm spatial resolution for targets embedded in homogenous turbid media .
In theory, the principles of ADFI can also be applied to time-resolved photon detection to potentially improve fluorescence lifetime property extraction. Based on the improved performance of ADFI over reflective imaging, it is expected that angular-domain fluorescence lifetime imaging (ADFLI) will provide images with greater spatial resolution of embedded sources compared to conventional FLI techniques. Furthermore, ADFLI should be more resistant to the temporal lengthening effects of depth on the surface calculation of lifetime since the mean differential path-length should be shorter.
3.1 System setup
The experimental setup based on the ADFLI configuration in the reflection mode is shown in Fig. 2(a) . In this mode, both the illumination source and the aligned detector system are located on the same side of the sample. In our system, a collimated beam from a pulsed-laser source was expanded to uniformly illuminate a sample containing two fluorescent targets. Then, the angular distribution of emitted photons from the embedded fluorescence targets was restricted by a Keplerian lens and pinhole system with an acceptance angle of 1.2° and, in some cases, was further restricted to 0.4° by an AFA. Finally, fluorescence emission was detected using a gateable intensified CCD camera, where time gating of the camera was synchronized to the pulsed laser source with a programmable delay.
A high power picosecond diode laser (780 nm wavelength) was employed for all experiments (PicoTA, PicoQuant GmbH and TOPTICA Photonics). A spectral filter (λ = 780 ± 5 nm) was used to block the background emission caused by the laser amplifier system. The diode laser had a pulse width of 100 ps (FWHM), a repetition rate of 80 MHz, and an average power of 500 mW. The laser beam (2.4 mm in diameter) was expanded into a collimated circular beam with a diameter of 1 cm. The AFA was aligned precisely at the same height as the light source to enable capture of the in-line fluorescent emission. A dichroic mirror was placed before the lenses of the Keplerian system to enable laser light delivery to the sample. The Keplerian lens and pinhole system was located between the sample and the AFA to transfer fluorescent emission from the sample to the AFA. Reflected laser light was rejected in the detection optical path by an emission bandpass filter (FL830-10, Thorlabs, NJ, USA) and a laser line long-pass filter (NT47-508, Edmund Optics Inc., NJ, USA) prior to reaching the camera.
Time-domain fluorescence was collected with a gateable intensified CCD camera (PicoStar HR, LaVision). A trigger delay unit (Delay Unit, LaVision, Germany) coupled to a synchronization signal from the laser driver was used to control a high-rate imager (HRI, Kentech Instruments, UK), which facilitated collection of CCD images from specific short-duration temporal windows. By performing a delay sweep, full fluorescence temporal point spread functions (FTPSFs) were constructed using a 500 ps gate width at 25-ps delay steps.
In all AFA optical imaging systems, including ADFLI, the light intensity at the output of the AFA is spatially modulated based on the micro-tunnel size and spacing as shown in Fig. 1(a). The line intensity profile has a sinusoidal pattern with the same periodicity as the micro-tunnel spacing, with signal peaks corresponding spatially to the center of the micro-tunnel openings. We attempted to remove the image artifacts by locating the upper envelope for each ADI line image and smoothing the resultant 2D ADI image in-row in order to eliminate the noise at each line profile in the intensity and lifetime maps. We used the “rloess” smoothing method in MATLAB (The Mathworks, Natick, MA), which employed local regression using weighted linear least squares and a 2nd degree polynomial model. Additional detail about artifact correction in angular domain imaging systems has been presented elsewhere .
The one dimensional linear array of micro-tunnels necessitated a scanning system for the capture of 2D images. We employed a computer-controlled z-axis stage to incrementally raise the sample between scans. One horizontal line image of the sample was taken through the AFA at each step and a final 2D image was assembled from the stacked line images. Hence, an entire region of the sample could be passed through the field of view of the AFA and imaged.
3.2 Phantom preparation
Three similar phantoms were constructed to explore both resolution (spatial and lifetime) and contrast (spatial and lifetime) of ADFLI. As shown in Figs. 2(b) and 2(c), phantoms were created by embedding two glass capillary tubes, each containing a different fluorophore with a distinct lifetime, at depths of 0, 1, or 2 mm, and separated laterally by 3 mm. The fluorescence sources were 20 μM solutions of 3,3-diethylthiatricarbocyanine iodide (DTTCI, lifetime ~1 ns) and indocyanine green (ICG, lifetime ~0.5 ns) embedded in a 1% Intralipid scattering medium (see Figs. 2(b) and 2(c)). DTTCI and ICG dyes were characterized separately using a Beckman DU-640 spectrometer. The absorption peak was found to be µa-DTTCI = 2.5 cm−1 at 758 nm for 20µM DTTCI and µa-ICG = 3.2 cm−1 at 772 nm for 20µM ICG. The absorption range for DTTCI and ICG at the excitation wavelength (λ = 780 ± 5 nm) was determined to be 1.8 cm−1 and 3 cm−1, respectively. Intralipid has optical scattering properties similar to tissue at visible and near infrared wavelengths. In addition, it has low optical absorption in the visible and near infrared regime [27,28]. Each phantom was imaged with the ADFLI system, with and without the AFA in place to investigate the performance of the AFA. As described in [27,28], considering the forward scattering property (g = 0.75) common in tissue-mimicking phantoms, the reduced scattering coefficient for 1% Intralipid is roughly 8-10 cm−1 in the near infrared, while the absorption coefficient is about 0.1-0.01 cm−1, which is two to three orders of magnitude smaller. The reciprocal value of reduced scattering coefficients yields the reduced mean free path (MFP'). We estimate that the minimum path for a fluorescent target at a depth of 2 mm will be about 4 MFP'. That is, photons at the excitation wavelength should travel two MFP' to reach the fluorophore and the emitted photons will travel a similar path, but in reverse, to reach the phantom surface.
3.3 Fluorescence lifetime map calculation
Fluorescence lifetime maps were created by fitting a single exponential decay, , where α is a weighting factor, τ is the fluorescent lifetime, and t 0 is a time shift parameter to account for delays in signal arrival (since no instrument response function was accounted for), to the down-slope of each FTPSF collected starting at 75% of the peak value of the FTPSF to the end of the collection window (typically about 30% of the peak value of the FTPSF). Fitting was performed with a Levenberg-Marquardt least squares algorithm applied in MATLAB. To avoid calculation of lifetime from pixels with minimal fluorescence signal, only FTPSFs whose sum was greater than 30% of the sum of the maximum intensity FTPSF in each image were fitted for lifetime. All other FTPSFs were assigned a lifetime of 0 ns.
3.4 Undersampling the fluorescence temporal response functions
To explore the feasibility of faster imaging times, FTPSFs from both ADFLI and lens and pinhole FLI were undersampled from 25-ps resolution to 50-, 75-, 100-, 125-, and 150-ps resolutions for all phantom images. Lifetime maps were then recalculated for each case to determine the effect of reduced temporal resolution on the lifetime estimates.
4. Results and discussion
4.1 Depth analysis
The intensity and lifetime maps obtained with and without the AFA from a pair of DTTCI and ICG sources at 0, 1, and 2 mm depths are shown in Fig. 3 . In addition, Fig. 4 shows the horizontal line intensity and lifetime profiles computed for each intensity and lifetime image for all depths to provide a clearer comparison between results with and without the AFA. The intensity maps in Fig. 3 were normalized in order to facilitate clearer comparisons between the images. However, the intensity line profiles in the Fig. 4 represent the absolute values as measured with the experimental setup in order to facilitate comparison of the sensitivity of the two methods as depth increased. The DTTCI and ICG sources are on the left and right side of each image, respectively. The dimensions of each image are 0.75 x 20 mm (height-x-width).
The images collected with the AFA revealed less broadening for fluorescent target lifetimes as the depth increased compared to images collected without the AFA. As a result, a clear improvement was observed in the spatial resolution of the fluorescence in the AFA technique, particularly for fluorescent sources at deeper depths. Furthermore, the lifetime contrast measured between the two fluorophore sources was significantly larger for the AFA technique for all source depths. The lifetime contrast ratios for AFA and No AFA images were 2.3 and 1.8 at 0 mm, 2.2 and 1.6 at 1 mm, and 2.0 and 1.4 at 2 mm, respectively. Lifetime of both DTTCI and ICG increased with depth of the fluorescence sources equally with and without the AFA. One source of this effect could be explained by dispersion of photons through tissue (which we did not attempt to account for) that would occur to a greater extent for deeper sources and would be roughly equivalent for both techniques. Theoretical analysis using the diffusion equation of light propagation in a homogeneous scattering media has suggested that a depth of 2 mm will result in an error in the lifetime estimate of approximately 0.01 ns . This level of error could be used to explain the increase in lifetime of the DTTCI inclusions with depth; however, the increase in ICG lifetime with depth was much larger. Therefore, the greater increase in lifetime of the ICG sources with depth may have been a phenomenon of a mixing of fluorescence signal from both sources, which would have occurred with greater propensity at deeper source positions. In other words, aside from an expected increase in measured lifetime with source depth, increasing source depth also increased the likelihood of fluorescence mixing from both tubes; wherein, the lifetime of the ICG would be expected to lengthen and the lifetime of DTTCI would be expected to shorten. However, since our procedure for estimating the lifetime used the later part of the FTPFS, the rapid signal decay from ICG may not have contributed to the shortening of the lifetime estimate for DTTCI, which is consistent with our data.
As shown in Fig. 4, the fluorescence emission intensity for the AFA case was approximately 5-fold lower than the No AFA case at 0 mm depth, but the ratio decreased to approximately 3-fold at 2 mm depth. In addition, the setup with the AFA had a reduced dependence of signal intensity on depth with a signal loss of approximately 50% (0 to 2 mm), where the setup lacking the AFA experienced a 3-fold drop in signal intensity for the same change in target depth. Greater noise was observed in the fluorescence cross-sections obtained with the AFA compared to No AFA. This owed to the restricted photon acceptance angle of the AFA, which permitted fewer photons to reach the detector. Additionally, the MCP gain of the camera system was increased to compensate for loss of signal and likely resulted in increased noise in the fluorescence images. Therefore, we hypothesize that the AFA system would be better suited to brighter fluorophores with higher extinction coefficients and quantum yields. Despite the reduced signal to noise ratio of the AFA images, it is clear from Fig. 4 that the AFA provided major improvements over the No AFA method in relation to resolution, contrast, and fluorescence lifetime contrast. For instance, it was very difficult to resolve the two separate sources at a depth 2 mm using the lifetime map collected without the AFA, while a clear separation was observed for the AFA lifetime map. Even with simple thresholding, the localization of different fluorophores was qualitatively easier with the AFA in place. Also, the lateral spatial resolution of the AFA system was expected to be uniform over the entire field of view due to the geometric and photon acceptance uniformity of the angular filter array.
The imaging results shown in Fig. 3 and Fig. 4 for the AFA system were captured when the sample was illuminated over a large area and detection was over a narrow line equivalent to the AFA aperture size. Since the illumination was over an area and only a small portion of the beam was co-aligned with the AFA, scattered light originating out of the detection plane is expected to contribute to a decrease in contrast. It is anticipated that shrinkage of the illumination area to a narrow line of light, co-linear with the AFA, will lead to a decrease in the background fluorescent signal and improved estimates of fluorescence intensity and lifetime.
4.2 Exposure time and effect of undersampling
For each scan line in ADFLI, the FTPSF was obtained by temporally scanning over 120 delay steps in 25 ps increments. For each point of the FTPSF, the camera exposure time was set to 600 ms with four times accumulation, which resulted in 2.4 sec acquisition time at each temporal delay step. Therefore, the time needed to acquire the FTPSF data with ADFLI was 288 s per horizontal line and collection of a 2D lifetime map depended directly on the number of lines used to create the 2D stack. Significant improvements to imaging speed could be realized by reducing the time to acquire each horizontal line. Following this approach, we tested the effect of undersampling the FTPSF at each line as a potential method to reduce acquisition time. As a result, we estimated the undersampling effect on the noise level of the lifetime signal. The noise was computed as the standard deviation of the lifetime estimates for each temporal sampling resolution. Specifically, we computedwhere N, x, and µ were the number of spatial points, lifetime estimates, and the smoothed lifetime estimates at computed at a 25-ps temporal sampling resolution. Figure 5(a) displays the effect of undersampling the FTPSF on the lifetime map for a phantom with targets at a depth of 1 mm collected with and without the AFA. Figure 5(b) displays the relationship between the standard deviation of the lifetime estimates and the temporal resolution used to collect the FTPSF with and without the AFA for ICG and DTTCI fluorophore agents at 1 mm depth.
Undersampling the FTPSF did not affect the spatial resolution of the lifetime maps for either the AFA or non-AFA systems; however, undersampling significantly increased the noise on the lifetime estimates. This was expected, since fluorescence lifetime quantification depends on the fidelity of each FTPSF, which is related to the number of samples used to estimate each FTPSF. Higher temporal sampling of each FTPSF is expected to provide better lifetime estimates, but at the cost of longer data acquisition time. The dependence of noise on temporal resolution was much less significant for lifetimes obtained from FTPSFs collected without the AFA.
In addition to FTPSF undersampling, imaging speed could be improved by incrementing the power of the excitation light and/or increasing the intensifier gain. In either case, integration times would be reduced, but with potentially undesirable consequences such as bleaching effects and/or introduction of additional electronic noise into the measurements. Undersampling the FTPSF provided an alternative method to speed up the image acquisition by a factor of 2 to 5 without increasing either the laser power or the intensifier gain and with relatively minor impact (~2-fold increase in noise for 5-fold decrease in acquisition time) on the noise in the lifetime estimates. Undersampling of the FTPSF may be well-suited to fluorophores with longer lifetimes.
In summary, phantom model tests demonstrated the usefulness of employing an angular filter array for fluorescent lifetime imaging to enhance the localization and lifetime determination of fluorescently-labeled targets. ADFLI has several advantages over other lifetime fluorescence imaging modalities, which include a larger field of view while preserving spatial image resolution, and better image contrast at tissue depths exceeding 1 mm. ADFLI could be used ultimately for the detection and sizing of tumor-targeted fluorescent agents at tissue depths significantly greater than conventional reflection-based imaging methods. Furthermore, precise and spatially-resolved fluorescent lifetime information could potentially improve quantification of the tumor microenvironment.
This project was funded by grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) to Drs. Jeffery J. L. Carson and Bozena Kaminska, and the Canadian Foundation for Innovation to Jeffery J. L. Carson. In addition, partial funding was provided by the Translational Breast Cancer Fellowship to Dr. Fartash Vasefi from the London Regional Cancer Program.
References and links
1. S. B. Bambot, J. R. Lakowicz, and G. Rao, “Potential applications of lifetime-based, phase-modulation fluorimetry in bioprocess and clinical monitoring,” Trends Biotechnol. 13(3), 106–115 (1995). [CrossRef]
4. J. R. Lakowicz, H. Szmacinski, K. Nowaczyk, and M. L. Johnson, “Fluorescence lifetime imaging of free and protein-bound NADH,” Proc. Natl. Acad. Sci. U.S.A. 89(4), 1271–1275 (1992). [CrossRef]
5. J. R. Lakowicz and H. Szmacinski, “Fluorescence lifetime-based sensing of pH, Ca2+, K+, and glucose,” Sens. Actuators B Chem. 11(1-3), 133–143 (1993). [CrossRef]
6. M. Y. Berezin, H. Lee, W. Akers, and S. Achilefu, “Near infrared dyes as lifetime solvatochromic probes for micropolarity measurements of biological systems,” Biophys. J. 93(8), 2892–2899 (2007). [CrossRef]
7. A. May, S. Bhaumik, S. S. Gambhir, C. Zhan, and S. Yazdanfar, “Whole-body, real-time preclinical imaging of quantum dot fluorescence with time-gated detection,” J. Biomed. Opt. 14(6), 060504 (2009). [CrossRef]
8. S. Gioux, S. J. Lomnes, H. S. Choi, and J. V. Frangioni, “Low-frequency wide-field fluorescence lifetime imaging using a high-power near-infrared light-emitting diode light source,” J. Biomed. Opt. 15(2), 026005 (2010). [CrossRef]
10. D. W. Piston, M. S. Kirby, H. Cheng, W. J. Lederer, and W. W. Webb, “Two-photon-excitation fluorescence imaging of three-dimensional calcium-ion activity,” Appl. Opt. 33(4), 662–669 (1994). [CrossRef]
11. A. Diaspro, Confocal and Two-Photon Microscopy: Foundations, Applications and Advances, (Wiley-VCH, New York, 2001) 101–125.
13. A. T. Kumar, J. Skoch, B. J. Bacskai, D. A. Boas, and A. K. Dunn, “Fluorescence-lifetime-based tomography for turbid media,” Opt. Lett. 30(24), 3347–3349 (2005). [CrossRef]
14. D. Elson, J. Requejo-Isidro, I. Munro, F. Reavell, J. Siegel, K. Suhling, P. Tadrous, R. Benninger, P. Lanigan, J. McGinty, C. Talbot, B. Treanor, S. Webb, A. Sandison, A. Wallace, D. Davis, J. Lever, M. Neil, D. Phillips, G. Stamp, and P. French, “Time-domain fluorescence lifetime imaging applied to biological tissue,” Photochem. Photobiol. Sci. 3(8), 795–801 (2004). [CrossRef]
15. D. Y. Paithankar, A. U. Chen, B. W. Pogue, M. S. Patterson, and E. M. Sevick-Muraca, “Imaging of fluorescent yield and lifetime from multiply scattered light reemitted from random media,” Appl. Opt. 36(10), 2260–2272 (1997). [CrossRef]
16. S. C. Davis, B. W. Pogue, R. Springett, C. Leussler, P. Mazurkewitz, S. B. Tuttle, S. L. Gibbs-Strauss, S. S. Jiang, H. Dehghani, and K. D. Paulsen, “Magnetic resonance-coupled fluorescence tomography scanner for molecular imaging of tissue,” Rev. Sci. Instrum. 79(6), 064302 (2008). [CrossRef]
18. D. S. Kepshire, S. C. Davis, H. Dehghani, K. D. Paulsen, and B. W. Pogue, “Subsurface diffuse optical tomography can localize absorber and fluorescent objects but recovered image sensitivity is nonlinear with depth,” Appl. Opt. 46(10), 1669–1678 (2007). [CrossRef]
19. A. T. Kumar, J. Skoch, B. J. Bacskai, D. A. Boas, and A. K. Dunn, “Fluorescence-lifetime-based tomography for turbid media,” Opt. Lett. 30(24), 3347–3349 (2005). [CrossRef]
21. F. Vasefi, B. S. Hung, B. Kaminska, G. H. Chapman, and J. J. L. Carson, “Angular domain optical imaging of turbid media using enhanced micro-tunnel filter arrays,” in Proceedings of SPIE 7369, 73691N (2009).
22. K. M. Tichauer, M. Najiminaini, F. Vasefi, T. Lee, B. Kaminska, and J. J. L. Carson, “Improved Lifetime Analysis Using Angular-Domain Fluorescence Imaging in a Tissue-Like Phantom,” in Biomedical Optics, OSA Technical Digest (CD) (Optical Society of America, 2010), paper BSuD36. [PubMed]
23. M. Najiminaini, F. Vasefi, B. Kaminska, G. H. Chapman, and J. J. L. Carson, “Macroscopic fluorescent lifetime imaging in turbid media using angular filter arrays,” in the 31st Annual International Conference of IEEE EMBS (Minneapolis, Minnesota, 2009), pp. 5364–5368.
24. F. Vasefi, E. Ng, B. Kaminska, G. H. Chapman, and J. J. L. Carson, “Angular domain fluorescent lifetime imaging in turbid media,” Proc. SPIE 7183, 71830I (2009). [CrossRef]
25. F. Vasefi, B. Kaminska, G. H. Chapman, and J. J. L. Carson, “Image contrast enhancement in angular domain optical imaging of turbid media,” Opt. Express 16(26), 21492–21504 (2008). [CrossRef]
26. F. Vasefi, A. Akhbardeh, M. Najiminaini, B. Kaminska, G. H. Chapman, and J. J. Carson, “Correction of Artifacts in Angular Domain Imaging,” in Biomedical Optics, OSA Technical Digest (CD) (Optical Society of America, 2010), paper BSuD36.
27. Optical properties of “Intralipid™”, an aqueous suspension of lipid droplets [Online]. Available: http://omlc.ogi.edu/spectra/intralipid/index.html.
28. H. J. van Staveren, C. J. Moes, J. van Marie, S. A. Prahl, and M. J. van Gemert, “Light scattering in Intralipid-10% in the wavelength range of 400-1100 nm,” Appl. Opt. 30(31), 4507–4514 (1991). [CrossRef]
29. S. H. Han, S. Farshchi-Heydari, and D. J. Hall, “Analysis of the fluorescence temporal point-spread function in a turbid medium and its application to optical imaging,” J. Biomed. Opt. 13(6), 064038 (2008). [CrossRef]