Open Access
Add to BibSonomy
Abstract
Wide field fluorescence microscopy is the most commonly employed fluorescence imaging modality. However, a major drawback of wide field imaging is the very limited imaging depth in scattering samples. By experimentally varying the control of illumination, we found that the optimized illumination profile can lead to large contrast improvement for imaging at a depth beyond four scattering path lengths. At such imaging depth, we found that the achieved image signal-to-noise ratio can rival that of confocal measurement. As the employed illumination control is very simple, the method can be broadly applied to a wide variety of wide field fluorescence imaging systems.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Thanks to the development of genetic and synthetic fluorescence indicators, fluorescence imaging [1,2] has profoundly transformed life science and medical research. Compared with other imaging technologies, fluorescence imaging has its key advantages in sensitivity, resolution and functionality. Owing to the background-free measurement, the large molecular absorption cross-section and the high quantum efficiency, fluorescence imaging offers the best sensitivity among all molecular imaging modalities. Single molecules [3] can be reliably detected through fluorescence measurements. Imaging resolution beyond the diffraction limit has also been enabled for fluorescence imaging through either nonlinearity or photo-switchable fluorophores [4–7]. Thus, protein distribution within single cells can be mapped at nanometer scale resolutions. Furthermore, the fusion of fluorophores with molecule binding groups has enabled a wide variety of function imaging probes, which allows tracking various biological signaling within live animals [8]. These advances have positioned fluorescence imaging as the dominant optical imaging modality for both structural and function imaging of biological systems.
The wonderful capabilities of fluorescence imaging rely on focusing light inside the sample [9–12]. Beyond cultured cell imaging, most applications require the propagation of light through samples of inhomogeneous refractive index distribution. The spatiotemporally random refractive index profile leads to optical wavefront distortion, manifested as light scattering. As a result, the controllable focus signal decays exponentially as a function of propagation depth, commonly known as Beer-Lambert law.
Wide field fluorescence microscopy remains the most widely used fluorescence imaging modality for its simplicity, convenience and availability. Typically, epi-detection was employed, in which the illumination source is coupled to the objective lens through a dichroic beam splitter. The fluorescence emission is collected by the same objective lens and imaged onto a CCD or CMOS camera for parallel detection. The simplicity comes with two drawbacks. One is the out-of-focus fluorescence signal. For imaging deep inside scattering samples, the fluorescence emission above and below the focal plane will produce a huge background. Although the uniform and slowly varying background profile can be computationally removed from the image, the photon shot-noise associated with the background will contaminate the true image and reduce the achievable signal-to-noise ratio (SNR). The second problem is the inefficient usage of fluorescence emission in scattering samples. As the ballistic component decays exponentially (Beer-Lambert law), only a small fraction of the emitted fluorescence signal contributes to the image information. The rest that reaches the sensor plane is scattered to other spatial modes, which leads to increased background noise [11,12].
Various solutions have been proposed and developed to enable fluorescence imaging in scattering samples. In general, these methods aim to solve the following two problems, the reduction of the out-of-focus background and the more efficient usage of the in-focus light generated fluorescence signals. For example, light-sheet imaging [13–19] has achieved great success for imaging of complex biological systems. The sheet illumination on the image focal plane can greatly reduce the out-of-focus fluorescence background and in addition avoid unnecessary photobleaching and phototoxicity. The large format parallel detection also allows a huge reduction of the illumination peak intensity, which is greatly desired for imaging living biological systems. Despite the greatly improved illumination configuration, the detection in light-sheet imaging remains the same as common wide field microscopy. Only the ballistic component contributes to the information content of the image and the scattered emission generates background and its associated shot noise. Multiphoton (mostly two-photon) excited fluorescence imaging solves both problems nicely [20–27]. The high order excitation inherently confines the fluorescence excitation to the laser focus. Therefore, all the excited fluorescence (ballistic and scattered) signal can be efficiently collected by large etendue optics and detected by a large area large access solid angle GaAsP photomultiplier tube. Moreover, the usage of NIR laser inherently offers deep penetration for its long scattering path length and negligible absorption in biological samples. The combined effect of out-of-focus suppression, efficient usage of fluorescence and the inherent longer excitation wavelength makes multiphoton fluorescence imaging the ideal choice for fluorescence imaging in scattering samples. The drawback is that the required high peak intensity of femtosecond laser pulses and the low excitation duty cycle (∼10−5 for two-photon) may lead to phototoxicity and photobleaching, especially for high-speed repetitive imaging.
With these established imaging tools considered, wide field imaging has its advantages in simplicity, compactness and availability. For these reasons, wide field imaging is also employed for miniature imaging systems such as head-mounted calcium imaging system for neuroscience applications and endoscopy system for medical applications [28–31]. In regard to phototoxicity and photobleaching, wide field imaging is actually rather gentle on the sample even with its out-of-focus excitation considered, as the peak intensity on the sample is very low and both phototoxicity and photobleaching are highly sensitive (nonlinear dependence) on peak intensity. In this work, we explore using wide field microscopy for imaging scattering samples. Specifically, we aim to find out if there is any simple way to further improve its deep imaging performance. The rationale for this study is that wide field microscope is such a widely available tool. So, if any improvement in performance is possible, it may broadly benefit many applications in biology and medicine.
2. Methods
To explore various imaging conditions, we developed a versatile system (Fig. 1) that allows the accurate control of illumination area, numerical aperture (NA) and angle and also allows a direct comparison between wide field imaging and confocal imaging. The system is a wide field fluorescence microscope with controls on laser illumination. By adding or removing a 5x beam expander, we can control the excitation NA. For the case of low NA excitation, we can translate the laser spot on the objective lens pupil by adjusting the position of a mirror supported on a precision linear stage. In this way, we can control the angle of incidence for the excitation beam. The galvo scanning allows us to synthesize an incoherent illumination area with well-defined transverse dimension. The addition of a diffuser in combination with 2-D galvo scanning allows us to mimic the typical illumination profile provided by an incoherent fluorescence light source. Finally, synchronizing the galvo scanning with the reading of individual pixels on the camera allows us to implement confocal detection in the same system for a direct comparison of imaging the same location.
Fig. 1. Imaging system: Laser, 488 nm diode laser; ND filter, variable neutral density filter wheel (OD 0-4); Beam expander, 5x beam expansion; Galvo, high-surface quality low-jitter xy non-resonant galvo scanner; L1 (f = 167 mm) and L2 (f = 300 mm), achromatic telecentric relay lenses; Dichroic, long-pass 488 nm reflection beam splitter; Tube lens, Nikon f = 200 mm tube lens; BP, band-pass filter for green emission; Camera, sCMOS camera.
The scattering fluorescence sample was made by dispersing 144 µL non-fluorescent 1 µm diameter polystyrene beads solution (weight 2.1% solid) and 144 µL 1 µm diameter green fluorescence beads (weight 2.6% solid) into 712 µL agar (weight 2% in water). Both the fluorophores and the scattering particles were randomly distributed. The measured scattering path length at the emission wavelength was ∼70 µm. For all imaging reported in this work, the imaging depth was fixed at 300 µm inside the sample (depth slightly above four scattering path lengths).
3. Results
Our goal is to optimize the imaging condition to achieve better performance for imaging thick scattering samples. For imaging thin and highly transparent sample, illumination field-of-view (FOV) and imaging performance (contrast, SNR) are unrelated. However, scattering can introduce signal crosstalk and therefore noise crosstalk. Thus, one hypothesis for imaging thick scattering samples is that a large illumination FOV could lead to increased background and background noise.
To test this hypothesis, we imaged the same sample using four different illumination methods (Fig. 2). First, we utilized low NA excitation in which the laser spot FWHM at the objective back focal plane can only support an NA of 0.13. To synthesize a spatially incoherent illumination profile, we raster scanned the laser spot over a 30 × 30 µm2 area. The raster scan period and camera exposure time were set identical. With such illumination, the individual 1 µm fluorescence beads could be clearly resolved albeit with a large background. For image processing, we first normalized the image to its maximum value. Next, we performed a 2D low-pass filtering to the image to represent its fluorescence background. We then subtracted this background from the original raw image. Alternatively, we can also directly perform 2D high-pass filtering to suppress background. With the image normalization and background subtraction, the resulting image magnitude is similar to the image contrast (see color bar value in Fig. 2). Individual 1 µm fluorescence beads can be clearly resolved with ∼3-4% contrast. Second, we flipped the 5x beam expander into the laser path to increase the illumination NA (NA 0.8 with a filling factor of 0.81). With the same 30 × 30 µm2 illumination area, the obtained image [Fig. 2(b)] showed similar contrast as that obtained in low NA illumination [Fig. 2(a)]. Third, we tried to mimic the illumination with an extended incoherent spatial light source (e.g. a lamp) by adding a diffuser. Again, the laser beam was expanded by the 5x beam expander. In addition, we commanded the galvo scanner to visit ∼500 × 500 µm2 area. With such an unconstrained area illumination, the image contrast [Fig. 2(c)] was drastically reduced. Individual 1 µm beads are barely visible. Fourth, we removed the diffuser and used the high NA illumination with 500 × 500 µm2 illumination area, which produced similar image contrast [Fig. 2(d)] as that based on the diffuser [Fig. 2(c)].
Fig. 2. Imaging the same sample using four different illumination conditions. The images are normalized with background subtracted. (a) Low NA with 30 × 30 µm2 illumination area. (b) High NA with 30 × 30 µm2 illumination area. (c) Diffuser with 500 × 500 µm2 illumination area. (d) High NA with 500 × 500 µm2 illumination area.
Through this experimental comparison, we found that the constrained illumination can drastically (∼4x) improve the image contrast compared to a large FOV illumination. The illumination NA, however, has little effect on the contrast or background level.
Next, we aimed to study the dependence of image contrast on the illumination area. Specifically, we employed the low NA beam and varied the galvo scan range (30, 60, 120, 500 µm). As expected, a large illumination area leads to reduced contrast (Fig. 3). It is worth noting that increasing the illumination area from 30 × 30 µm2 to 60 × 60 µm2 only causes minor contrast difference.
Fig. 3. The image contrast dependence on the illumination area with low NA illumination. (a) 30 × 30 µm2. (b) 60 × 60 µm2. (c) 120 × 120 µm2. (d) 500 × 500 µm2.
These observations can be explained by considering the size mismatch between the scattering induced illumination energy spread and the fluorescence background. Let’s assume that the fluorophores are randomly distributed inside the same. Given the same amount of illumination energy (IE), the amount of background fluorescence energy (BE) is constant regardless of the illumination area (IA). Therefore, the average background signal recorded on the camera is proportional to ${BE}/({BS \otimes IA}$), where BS is the inherent area spread of the background. Similarly, the excitation intensity at the focal plane is proportional to $ {IE}/({ {ES \otimes IA}} )$, where ES is the inherent area spread of the excitation energy as a result of scattering. Thus, the obtained image contrast is proportional to $({ {BS \otimes IA}} )$/$( {ES \otimes IA}$). The background (e.g. see Fig. 4) has an inherent spread over ∼ 200 µm (BS). From the variation of image contrast in Fig. 3, the inherent excitation energy spread should be around ∼60 µm (ES). For illumination area less than 60 × 60 µm2, the effective illumination area ($ {ES \otimes IA}$) remains the similar value as ES and the effective background spread ($ {BS \otimes IA}$) remains the similar value as BS. Therefore, the contrast remains similar. For the illumination area of 120 × 120 µm2, $ {ES \otimes IA}$ is increased. However, $ {BS \otimes IA}$ remains similar value. As a result, the image contrast is decreased. For the illumination area of 500 × 500 µm2, both ${ES \otimes IA}$ and ${BS \otimes IA}$ greatly exceed their native value (ES and BS). However, the increase on the illumination profile was more significant due to its smaller inherent spread. As a result, the image contrast was further decreased. Overall, the consideration of the inherent spread for the background signal and the illumination energy and their variation through convolution allows us to appreciate the benefit of constrained illumination for wide-field imaging in scattering samples.
Fig. 4. (a-c) Normalized and background subtracted images recorded with incident illumination angle of 0, 16 and 30 degree, respectively. The illumination area is 30 × 30 µm2. (d-f) The corresponding normalized images without background subtraction.
Next, we aim to study the effect incident illumination angle. For this study, we utilized the low NA illumination and varied the laser spot location on the back focal plane of the objective lens by translating the mirror on the linear stage in Fig. 1. In this way, we varied the incident angle from 0 to 30 degrees (Fig. 4). Using a large incident angle could shift the out-of-focus fluorescence way from the imaging axis and thus lower the background at the center of the image. However, at a large incident angle, the illumination beam path in the scattering samples is even longer and therefore suffers more from scattering induced energy spreading. The combined effect is that the image contrast remains similar.
As the out-of-focus fluorescence contributed substantially to the image background, it would be interesting to compare wide field microscopy with imaging methods that can reject out-of-focus signals. A well-established standard is confocal microscopy [11]. To make a direct comparison at the same imaging location, we employed the high NA illumination and synchronized the galvo scanning with the individual pixel readout of the detection camera, which effectively implemented confocal scanning and detection.
As expected, confocal imaging produced images of much greater contrast (∼120%, Fig. 5). However, a key factor to consider for practical application is the image SNR, which depends on both the contrast and the total signal strength (accumulated photon number). Given the same amount of illumination energy, will confocal outperform wide field imaging in SNR?
Fig. 5. Normalized and background subtracted images recorded by confocal microscopy (a) and wide field microscopy (b, c) with the same illumination power and time. (c) The zoomed-in view of the dotted box in b.
Suppose the total illumination energy is E and the total number of image pixel is N. For confocal imaging, the energy utilized for each pixel is therefore E/N. The background signal arriving at the detector (only one pixel for confocal) is bE/N, where b is a constant. The contrast for confocal imaging is C1 and therefore the useful image signal is C1bE/N. Thus, the total light collected by the detector is (1+ C1) bE/N. At the shot noise limited condition, the detection noise is $\sqrt {({1 + {C_1}} )\; bE/N} $. The SNR is therefore $\sqrt {bEC_1^2/({1 + {C_1}} )/N} $. For wide field imaging, the total illumination energy is simultaneously delivered to all pixels. The energy utilized for each pixel is E/N. The background generated by the illumination of one pixel arriving at the camera sensor is bE/N per pixel. As all N pixels are illuminated at the same time, their background (spread over several hundred of microns) will accumulate on each pixel (background crosstalk). As a result, the accumulated background is bE on each pixel, which is much worse than the case of confocal detection. Suppose the wide field imaging contrast is ${C_2}$ and therefore the useful image signal is C2bE. ${C_2}$is a small number and thus the shot noise is ∼$\sqrt {bE} $. The SNR for wide field imaging is therefore $\sqrt {bEC_2^2} $. The ratio of the SNR of wide field imaging to the SNR of confocal imaging is therefore $\sqrt {N({1 + {C_1}} )C_2^2/C_1^2} $. For 100 × 100 pixels ($N = 10000,\; {C_1} = 120\%,\; {C_2} = 3\%$), the ratio is 3.7, which suggests that wide field imaging can in fact achieve comparable measurement SNR as that of confocal imaging.
To test this conclusion, we employed the same illumination energy (same power, same total integration time) for both confocal microscopy and wide field microscopy with constrained illumination for recording 100 × 100 pixels (Fig. 5). The SNR is manifested as the pixel brightness variation on the fluorescent beads. Despite its lower contrast [Fig. 5(b)], the image SNR of wide field measurement is comparable to that of confocal measurement. It is worth noting that the surrounding area near the illumination area [dotted box in Fig. 5(b)] is also recorded with decent SNR. Thus, given the same illumination energy, wide field imaging not only produces comparable image SNR but also achieves greater measurement area (information content).
4. Discussion
Through experimental measurement, we found that constrained illumination can effectively improve the image contrast in wide field imaging of scattering samples. Based on the measurement and the analysis of the contrast dependence on the illumination area (Fig. 3), the image contrast is optimized when the illumination area is comparable or less than the inherent illumination energy spread due to random scattering. Therefore, care must be taken for the design of wide field illumination system for imaging scattering samples, especially for the case of incoherent light sources based systems. Proper spatial constrain (filtering) on the light source (e.g. an aperture imaged onto the sample plane) can easily implement the constrained illumination. In comparison, the NA and the incident angle of illumination do not significantly alter the image contrast.
A limitation of this contrast gain is that the FOV is limited (e.g. ∼ 60 µm in Fig. 3). Such a FOV can benefit the study of single cells, in vivo flow cytometry and micro-endoscope. For imaging multiple targets distributed over a much larger FOV, patterned illumination (sparse and confined illumination for each target) can be applied to achieve contrast gain.
The contrast gain reported in this work is due to the mismatch between the excitation energy spread and the background spread. The excitation energy spread is a result of random scattering while the background spread is a result of out-of-focus fluorescence and random scattering. For biological tissue, scattering is often dominated by forward scattering. The average of the cosine of the scattering angle is near ∼0.9 for a wide range of biological tissue. Therefore the excitation energy has a limited spread within the imaging depth range. In comparison, the fluorescence emission is statistically omnidirectional. In clear samples, the out-of-focus contribution has an angular spread limited by the numerical aperture of the detection objective lens. Moreover, the out-of-focus contribution comes from the whole transport length of the illumination light inside samples, which is often much greater than the imaging depth range. In thick scattering samples, the omnidirectional emission combined with random scattering can further increase the background spread. Together, these factors contributed to the mismatch between the excitation energy spread and the background spread.
Confocal imaging can significantly improve the image contrast. However, better contrast does not necessarily imply better image SNR. The photon number collected on each image pixel also matters. Although wide field imaging suffers from strong background crosstalk (the background due to the illumination of one point can spread across a large number of pixels), the simultaneous parallel image collection also greatly boosts the signal level. This is made possible by the random scattering induced illumination energy spread. Even if we try to focus the laser beam onto just one spot, the majority (non-ballistic components) of the energy is actually spread around the ballistic focus to simultaneously light up the surrounding area. For confocal detection, the useful signal excited from the surrounding area is fully rejected. With wide field detection, these signals can be efficiently accumulated. So, depending on the imaging area, wide field imaging can actually achieve comparable image SNR despite its lower contrast. For applications in biology and medicine, wide field imaging also has its advantages in much reduced peak intensity in comparison with confocal imaging for reduced photobleaching and phototoxicity.
The measurement and analysis in this work is based on a uniformly and randomly distributed fluorophores, which is in fact valid for a wide variety of applications (e.g. glial cells and neurons in brain tissue, immune cells in lymph nodes). The image processing relies on our prior knowledge of the spatial profile of the fluorophores. Without such prior knowledge, we cannot use the simple background subtraction method. More sophisticated 3D image deconvolution [32–34] should be applied.
The comparison between confocal and wide field microscopy also inspires the favorable consideration on light sheet imaging. As a wide field recording method, light sheet imaging can effectively reduce the out-of-focus contribution, which can greatly improve both the contrast and image SNR. One constrain on choosing light sheet imaging over simple wide field imaging is the space and geometry constrain. Not all samples are suitable for imaging between two orthogonally positioned objectives. The single objective lens based light-sheet also requires high NA. For space and NA constrained applications [35] (e.g. endoscopy or miniature imaging systems), simple wide field imaging is still preferred.
5. Conclusion
In summary, we evaluated various illumination schemes for further improving the image contrast of wide field imaging of scattering samples. A simple constrain on the illumination area can lead to significant contrast gain. Moreover, we show through analysis and measurement that wide field imaging despite its low contrast can actually achieve comparable image SNR as that of confocal imaging in scattering samples. These results will be valuable to further improve the wide field imaging performance for scattering samples through both simple illumination control and better background removal algorithms. As wide field fluorescence imaging remains the most widely employed fluorescence imaging modality, these findings will broadly benefit the applications in life science and medicine, especially for the cases where light-sheet imaging cannot be applied.
Funding
National Institute of Neurological Disorders and Stroke (U01NS094341, U01NS107689).
Acknowledgments
Z.-Y.C. acknowledges the support from the Ph.D. abroad visiting scholar program of the Nanchang University. S. S. acknowledges the financial support from the program of China Scholarships Council. M.C. acknowledges the scientific equipment support from HHMI.
Disclosures
The authors declare no conflicts of interest.
References
1. B. N. G. Giepmans, S. R. Adams, M. H. Ellisman, and R. Y. Tsien, “Review - The fluorescent toolbox for assessing protein location and function,” Science 312(5771), 217–224 (2006). [CrossRef]
2. R. Y. Tsien, “The green fluorescent protein,” Annu. Rev. Biochem. 67(1), 509–544 (1998). [CrossRef]
3. R. M. Dickson, A. B. Cubitt, R. Y. Tsien, and W. E. Moerner, “On/off blinking and switching behaviour of single molecules of green fluorescent protein,” Nature 388(6640), 355–358 (1997). [CrossRef]
4. S. W. Hell and J. Wichmann, “Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy,” Opt. Lett. 19(11), 780–782 (1994). [CrossRef]
5. M. J. Rust, M. Bates, and X. Zhuang, “Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM),” Nat. Methods 3(10), 793–796 (2006). [CrossRef]
6. E. Betzig, G. H. Patterson, R. Sougrat, O. W. Lindwasser, S. Olenych, J. S. Bonifacino, M. W. Davidson, J. Lippincott-Schwartz, and H. F. Hess, “Imaging intracellular fluorescent proteins at nanometer resolution,” Science 313(5793), 1642–1645 (2006). [CrossRef]
7. M. G. Gustafsson, “Nonlinear structured-illumination microscopy: wide-field fluorescence imaging with theoretically unlimited resolution,” Proc. Natl. Acad. Sci. U. S. A. 102(37), 13081–13086 (2005). [CrossRef]
8. T.-W. Chen, T. J. Wardill, Y. Sun, S. R. Pulver, S. L. Renninger, A. Baohan, E. R. Schreiter, R. A. Kerr, M. B. Orger, V. Jayaraman, L. L. Looger, K. Svoboda, and D. S. Kim, “Ultrasensitive fluorescent proteins for imaging neuronal activity,” Nature 499(7458), 295–300 (2013). [CrossRef]
9. L. V. Wang, “Multiscale photoacoustic microscopy and computed tomography,” Nat. Photonics 3(9), 503–509 (2009). [CrossRef]
10. L. V. Wang and H.-I. Wu, Biomedical optics : principles and imaging (Wiley-Interscience, Hoboken, New Jersey, 2007).
11. J. Pawley, Handbook of Biological Confocal Microscopy, 3rd ed. (SpringerUS, 2006).
12. T. Vo-Dinh, Biomedical photonics handbook (CRC press, New York, 2003).
13. P. J. Keller, A. D. Schmidt, J. Wittbrodt, and E. H. Stelzer, “Reconstruction of zebrafish early embryonic development by scanned light sheet microscopy,” Science 322(5904), 1065–1069 (2008). [CrossRef]
14. J. Huisken, J. Swoger, F. Del Bene, J. Wittbrodt, and E. H. K. Stelzer, “Optical sectioning deep inside live embryos by selective plane illumination microscopy,” Science 305(5686), 1007–1009 (2004). [CrossRef]
15. M. B. Bouchard, V. Voleti, C. S. Mendes, C. Lacefield, W. B. Grueber, R. S. Mann, R. M. Bruno, and E. M. C. Hillman, “Swept confocally-aligned planar excitation (SCAPE) microscopy for high-speed volumetric imaging of behaving organisms,” Nat. Photonics 9(2), 113–119 (2015). [CrossRef]
16. T. F. Holekamp, D. Turaga, and T. E. Holy, “Fast three-dimensional fluorescence imaging of activity in neural populations by objective-coupled planar illumination microscopy,” Neuron 57(5), 661–672 (2008). [CrossRef]
17. C. Dunsby, “Optically sectioned imaging by oblique plane microscopy,” Opt. Express 16(25), 20306–20316 (2008). [CrossRef]
18. B.-C. Chen, W. R. Legant, K. Wang, L. Shao, D. E. Milkie, M. W. Davidson, C. Janetopoulos, X. S. Wu, J. A. Hammer III, Z. Liu, B. P. English, Y. Mimori-Kiyosue, D. P. Romero, A. T. Ritter, J. Lippincott-Schwartz, L. Fritz-Laylin, R. D. Mullins, D. M. Mitchell, J. N. Bembenek, A.-C. Reymann, R. Boehme, S. W. Grill, J. T. Wang, G. Seydoux, U. S. Tulu, D. P. Kiehart, and E. Betzig, “Lattice light-sheet microscopy: Imaging molecules to embryos at high spatiotemporal resolution,” Science 346(6208), 1257998 (2014). [CrossRef]
19. L. Gao, L. Shao, C. D. Higgins, J. S. Poulton, M. Peifer, M. W. Davidson, X. Wu, B. Goldstein, and E. Betzig, “Noninvasive Imaging beyond the Diffraction Limit of 3D Dynamics in Thickly Fluorescent Specimens,” Cell 151(6), 1370–1385 (2012). [CrossRef]
20. P. Theer and W. Denk, “On the fundamental imaging-depth limit in two-photon microscopy,” J. Opt. Soc. Am. A 23(12), 3139–3149 (2006). [CrossRef]
21. P. Theer, M. T. Hasan, and W. Denk, “Two-photon imaging to a depth of 1000 µm in living brains by use of a Ti : Al2O3 regenerative amplifier,” Opt. Lett. 28(12), 1022–1024 (2003). [CrossRef]
22. F. Helmchen, K. Svoboda, W. Denk, and D. W. Tank, “In vivo dendritic calcium dynamics in deep-layer cortical pyramidal neurons,” Nat. Neurosci. 2(11), 989–996 (1999). [CrossRef]
23. D. Kleinfeld, P. P. Mitra, F. Helmchen, and W. Denk, “Fluctuations and stimulus-induced changes in blood flow observed in individual capillaries in layers 2 through 4 of rat neocortex,” Proc. Natl. Acad. Sci. 95(26), 15741–15746 (1998). [CrossRef]
24. K. Svoboda, W. Denk, D. Kleinfeld, and D. W. Tank, “In vivo dendritic calcium dynamics in neocortical pyramidal neurons,” Nature 385(6612), 161–165 (1997). [CrossRef]
25. W. Denk and K. Svoboda, “Photon upmanship: Why multiphoton imaging is more than a gimmick,” Neuron 18(3), 351–357 (1997). [CrossRef]
26. R. Yuste and W. Denk, “Dendritic spines as basic functional units of neuronal integration,” Nature 375(6533), 682–684 (1995). [CrossRef]
27. N. G. Horton, K. Wang, D. Kobat, C. G. Clark, F. W. Wise, C. B. Schaffer, and C. Xu, “In vivo three-photon microscopy of subcortical structures within an intact mouse brain,” Nat. Photonics 7(3), 205–209 (2013). [CrossRef]
28. R. P. Barretto, T. H. Ko, J. C. Jung, T. J. Wang, G. Capps, A. C. Waters, Y. Ziv, A. Attardo, L. Recht, and M. J. Schnitzer, “Time-lapse imaging of disease progression in deep brain areas using fluorescence microendoscopy,” Nat. Med. 17(2), 223–228 (2011). [CrossRef]
29. B. A. Flusberg, J. C. Jung, E. D. Cocker, E. P. Anderson, and M. J. Schnitzer, “In vivo brain imaging using a portable 3.9 gram two-photon fluorescence microendoscope,” Opt. Lett. 30(17), 2272–2274 (2005). [CrossRef]
30. J. C. Jung, A. D. Mehta, E. Aksay, R. Stepnoski, and M. J. Schnitzer, “In vivo mammalian brain imaging using one-and two-photon fluorescence microendoscopy,” J. Neurophysiol. 92(5), 3121–3133 (2004). [CrossRef]
31. D. Zaak, M. Kriegmair, H. Stepp, H. Stepp, R. Baumgartner, R. Oberneder, P. Schneede, S. Corvin, D. Frimberger, and R. Knüchel, “Endoscopic detection of transitional cell carcinoma with 5-aminolevulinic acid: results of 1012 fluorescence endoscopies,” Urology 57(4), 690–694 (2001). [CrossRef]
32. J. R. Swedlow, J. W. Sedat, and D. A. Agard, “Deconvolution in optical microscopy,” in Deconvolution of images and spectra (2nd ed.), (Academic Press, Inc., 1996), 284–309.
33. D. A. Agard, Y. Hiraoka, P. Shaw, and J. W. Sedat, “Fluorescence microscopy in three dimensions,” in Methods in cell biology (Elsevier, 1989), pp. 353–377.
34. D. A. Agard, “Optical sectioning microscopy: cellular architecture in three dimensions,” Annu. Rev. Biophys. Bioeng. 13(1), 191–219 (1984). [CrossRef]
35. D. Aharoni, B. S. Khakh, A. J. Silva, and P. Golshani, “All the light that we can see: a new era in miniaturized microscopy,” Nat. Methods 16(1), 11–13 (2019). [CrossRef]
