Adaptive Optics Confocal Fluorescence Microscopy with Direct ...
Adaptive Optics Confocal Fluorescence Microscopy with Direct Wavefront Sensing for Brain Tissue Imaging Xiaodong Tao*a, Bautista Fernandez a, Diana C. Chenb, Oscar Azucena a, Min Fu c, Yi Zuo c and Joel Kubby a a Jack Baskin School of Engineering, Univ. of California, Santa Cruz, 1156 High St., Santa Cruz, CA, USA 95064; b Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, CA, USA 94550; c Molecular, Cell, and Developmental Biology, Univ. of California, Santa Cruz, 1156 High St., CA, USA 95064 ABSTRACT Recently, there has been a growing interest in deep tissue imaging for the study of neurons. Unfortunately, because of the inhomogeneous refractive index of the tissue, the aberrations degrade the resolution and brightness of the final image. In this paper, we describe an adaptive optics confocal fluorescence microscope (AOCFM) which can correct aberrations based on direct wavefront measurements using a point source reference beacon and a Shack-Hartmann Wavefront Sensor (SHWS). Mouse brain tissues with different thicknesses are tested. After correction, both the signal intensity and contrast of the image are improved. Keywords: Adaptive Optics, confocal microscopy, wavefront sensor, wavefront correction, deep tissue imaging
1. INTRODUCTION Fluorescence microscopy has been widely used for the study of the neuron, such as monitoring biochemical signaling pathways [1], visualization of neuronal dynamics [2], determinations of cellular morphologies [3] and mapping protein distribution in cells [4]. For imaging of the long dendrite and tiny dendritic spines of the neuron, high-resolution fluorescence imaging deep within brain tissue is required. Although some objectives are equipped with a correction collar for compensation of the spherical aberration from the cover slide, the high numerical objectives that are used still suffer aberration from the inhomogeneous refractive index of the tissue and the refractive index mismatch between the sample and mounting medium [5]. For in-vivo imaging, the non-uniform skull thickness may also cause significant spherical aberrations [6]. Compared with the wide-field microscope, the confocal microscope suffers from more aberration because both the excitation and emission light are distorted when they go though the sample [7]. The distortion in excitation light will decrease the intensity of the fluorescent emission light. Using a higher power laser to compensate this effect will cause phototoxicity and photobleaching [8]. Both of these distortions also reduce the resolution of the system. Moreover, the distortion increases with the imaging depth. A similar issue can be found in astronomy where the aberration induced by atmospheric turbulence degrades the final image from the telescope [9]. In order to compensate this dynamic aberration, the wavefront distortion is measured directly by a wavefront sensor. An active optical component, such as deformable mirror (DM) or spatial light modulator (SLM), corrects the aberration in a feed back loop [9]. However, wavefront sensors often need a reference point source, such as laser guide star, which is difficult to realize in biological specimens. Because of this, most of the existing adaptive optics (AO) confocal microscopes developed so far are based on indirect methods of wavefront measurement which depends on processing of the final image [10,11,12,13]. In order to find the wavefront parameters, numerous iterations are required, which will cause photobleaching and limits the bandwidth for live imaging. Backscattered light has been used for direct wavefront sensing [14]. However it is highly dependent on the backscattering efficiency of the tissue. The complex procedure used for wavefront sensing makes it computationally intensive. *[email protected]; phone (831) 295-3957 MEMS Adaptive Optics V, edited by Scot S. Olivier, Thomas G. Bifano, Joel A. Kubby, Proc. of SPIE Vol. 7931, 79310L · © 2011 SPIE · CCC code: 0277-786X/11/$18 · doi: 10.1117/12.876524
Proc. of SPIE Vol. 7931 79310L-1
In this paper, we describe an adaptive optics confocal fluorescence microscope (AOCFM) which can correct aberrations based on direct wavefront measurements using a point source reference beacon and a Shack-Hartmann Wavefront Sensor (SHWS). In order to measure the wavefront directly, fluorescent microspheres with a diameter of 1µm can be injected into the mouse brain tissue. These microspheres were used as reference beacons for the adaptive AO system, similar to the artificial guide stars used in astronomical AO systems [15]. The microsphere is excited by a HeNe laser (633nm). The emission from the microsphere goes through a filter, which blocks the emission from the tissue, and is finally collected by a SHWS comprised of a lenslet array focused on a cooled CCD camera. The control signal for the deformable mirror is calculated based on the estimated wavefront error from the SHWS. In the experiment, mouse brain tissues with different thicknesses are placed on top of the microsphere reference beacon to test the proposed AO system. The neurons were marked by the transgenic expression of yellow fluorescent protein (YFP).
2. METHOD 2.1 Optical System Layout Fig.1. Shows the layout of the AOCFM used for brain tissue imaging. The system was designed and optimized using the optical design software (CODE V). A 60X water immersion objective with a numerical aperture of 1.2 was used (Olympus Microscope, Center Valley, PA) for imaging of the dendrites and spines in brain tissue. The optical system includes three telescope relay sub-systems. Lenses L1 and L2 image the exit pupil of objective onto the Y scanner. Lenses L3 and L4 relay the X scanner conjugate to the Y scanner. This design minimizes the movement of scanning beam at the exit pupil of objective and the emission light at the DM, which is important for accurate wavefrront measurement and correction. Lenses L2 and L3 also serve as scanning lenses. The current design is optimized for an optical scanning angle of 4.4 degree, which provides a field of view of 128 µm on the sample with a 60x objective. By changing the control signal of the scanners, the field of view can be easily adjusted. Lens L5 and L6 image the pupil of the X scanner onto the DM. Lenses L7 and L8 image the pupil of DM onto the wavefront sensor. The diameter of the effective aperture on the DM used in this design is 4 mm. The exit pupil of the objective is 7.2 mm. To match the aperture of the DM with the objective, the telescope L1 and L2 demagnifies the pupil from 7.2 mm to 4mm.
L10 L9
L8
L7 L6 L5
L4
L3
L2 L1
L: Lens
Figure 1. System Layout of the Adaptive Optics Confocal Microscope. A HeNe laser emits light at 633nm for excitation of the fluorescent reference beacons. A solid state laser emits light at 515 nm that excites yellow fluorescent protein (YFP) bred into the sample. F2 and F4 are excitation filters. Light emitted from the reference beacon is passed through filter F1 to the wavefront sensor. The dichroic beamsplitters DB2 separate the light from reference beacon and the sample. The dichroic beamspliter DB1 and DB3 are used to separate the excitation light and its associated back reflection. The fluorescent light emitted by the YFP is filtered by F3 and detected by the Photo-Multiplier Tube (PMT). The wavefront aberration is corrected by the Deformable Mirror (DM).
Proc. of SPIE Vol. 7931 79310L-2
Two laser channels are included in the system, a solid state laser (λ=515nm) and a helium neon laser (λ=633nm) used for confocal fluorescence imaging and wavefront sensing, respectively. The two channels share the same lightpath through the DM, relay lens, scanner and scanning lens and feed into an Olympus IX71 inverted microscope (Olympus Microscope, Center Valley, PA) from its side optical port. The focused beam is scanned on the sample in a raster pattern with a resonant scanner (SC-30, 16 kHz, 5°, Electro-Optics Products Corp) and a vertical scanner (GVS001, Thorlabs). The control signals for the scanners are generated by a data acquisition board (PCIe-6363, National Instruments Corporation). The emission light from both the microsphere and the sample are collected by an objective lens (Olympus 60x, 1.2 NA water immersion) and separated by a dichroic mirror (DB2). The light from the sample was focused on a pin hole by an achromatic lens L10. A GaAs photomultiplier tube (PMT) (H422-50, Hamamatsu) was used as a photon detector. The signal from the PMT is then fed to a frame-grabbing board (Helios eA/XA, Matrox Imaging). With the vertical and horizontal synchronized signal from the DAQ, the frame-grabbing board generates the raw image of 512x512 pixels at 30 frames per second. When the signal levels are low, more frames can be averaged to increase the signal to noise ratio. The experimental system was setup on an optical table with vibration isolation as shown in Fig. 2. The beam from the scanning system was fed into Olympus IX71 inverted microscope by a periscope. The size of the system without the microscope frame is about 48 inch ×28 inch.
Science Camera
OLYMPUS X61 Frame
PMT HeNe Laser
Solid State Laser
DM
SH Wavefront sensor
X Scanner Y Scanner Figure 2. Experimental setup of Adaptive Optics Confocal Microscope. The system was mounted on an optical table with vibration isolation.
2.2 Wavefront Sensing To correct the aberration induced by the specimen, accurate measurement of the wavefront becomes an important factor. SHWS has been widely used in astronomy and vision science, which consists of a lenslet array and a camera at its focal plane. By measuring the displacement of spots imaged by the lenslet array from their reference positions, the wavefront error can be estimated [16]. In the current design, the light from the microsphere is fed into the SHWS with 44×44 element lenslet array (AOA Inc., Cambridge, MA) and an electron multiplying CCD camera (Photometrics). The pitch of the lenslet array is 400 µm, which is equal to the pitch of the actuators in the DM. The SHWS requires a reference source to produce a spot image on the sensor. In astronomy, a real star or a laser guid star are often used as the reference source [9]. However, it is difficult to find one in the biological sample. In order to measure the wavefront directly, fluorescent microspheres with a diameter of 1µm can be imbedded in the sample as
Proc. of SPIE Vol. 7931 79310L-3
reference source (FluoSpheres crimson fluorescent microspheres, Invitrogen, Carlsbad, CA) [15]. In order to reduce the overlap between the biology sample and fluorescent microspheres on both the excitation and the emission side, the microspheres and dichroic filter are selected based on the spectrum of the fluorescent protein in the biology sample. In this study, the neuron in the brain tissue is YFP labeled. Its peak absorption and emission wavelength are 514 nm and 527 nm. A solid state laser ((λ=515 nm) was used as the excitation source. The spectrum profile is show in Fig. 3. A single-band laser filter set for YFP (LF514-A, Semrock, Inc.) is used in the system. The transmittance profile of the emission filter in the filter set is shown as a purple line in Fig. 3. The absorption and emission spectra profile of the microsphere is also shown in Fig. 3. The overlap between the emissions from YFP and microsphere is only 3% of the total emission power. The peak absorption and emission wavelength of the microsphere are 625 nm and 645 nm. A HeNe laser was used as the excitation source. An ultra-steep long-pass edge filter (LP02-633RU, Semrock, Inc.) is used as the emission filter which was mounted on the wavefront sensor. 83% of emission light from the microsphere passes though the filter. Only 2.7% of emission light from YFP goes through the filter. The emission filter for the YFP can block the light from the microsphere. By demixing the light path between confocal imaging and wavefront sensing channels, this design reduces the noise in wavefront measurement and blocks the light from the microsphere in the final confocal image. Before wavefront measurement, a SH reference image was captured when a collimated laser beam was added to the optical path after the DM. The position of the spots in the image from the wavefront sensor was detected using a crosscorrelation centroiding algorithm [17]. During the wavefront measurement, the wavefront slopes can be obtained from the displacement of spots in the image. Then a Fast Fourier Transform (FFT) reconstruction algorithm was implemented to obtain the real wavefront [16]. It can measure the wavefront induced from the sample, mounting medium, cover slide and misaligned optical components. Excitation source for microsphere (633nm)
Excitation source for YFP (515nm) 100 90
70
90 80 70
60
60
50
50
40
40
30
30
20
20
10
10
0 300
350
400
450
500
550
600
650
700
750
800
Relative Intensity (%)
80
Transmittance (%)
100
YFP Emission YFP Absorption Microsphere Emission Microsphere Absorption Emission Filter for Microsphere Emission Filter for YFP
0
Figure 3. The absorption/emission spectral profile and transmittance profile of the emission filters for YFP and fluorescent microsphere. The wavelengths of the excitation source for YFP and the microsphere are 515 nm and 633 nm, respectively.
2.3 Wavefront Compensation In order to correct the wavefront, a DM (Boston Micromachines) with 140 actuators and 3.5 µm of stroke is placed in the optical path, where it is conjugate to the exit pupil of the objective, the X and Y scanners and the wavefront sensor. Because the excitation and emission light shares the same path, the DM can correct the aberration on both paths. The effective aperture size of the DM used in this design is 4 mm, which is the same as the wavefront sensor. With the correction information from the wavefront sensor, a direct slope algorithm was applied to control the DM in a closed loop [18]. The influence matrix of the DM was calibrated first with the microsphere under the cover slide outside of the sample. For the real wavefront compensation, the slopes of the wavefront were measured from the displacement of the spots on the wavefront sensor. The control signal for the DM is then calculated by multiplying the pseudo inverse of the influence matrix of the DM and the slope measurement with a proportional gain of 0.4.
Proc. of SPIE Vol. 7931 79310L-4
3. EXPERIMENTAL RESULT 3.1 Sample Preparation To investigate the feasibility of the proposed system, a fixed brain slice from YFP-H line transgenic mouse was prepared. YFP is labeled on the cell body and protrusions of neurons. One micron diameter crimson fluorescent microspheres (Invitrogen, Carlsbad, CA) were deposited onto a glass slide and a cover plate for use as laser guide stars. Sample brain coronal sections of different thicknesses (15 μm, 50 μm and 100 μm) were cut with a microtome. The tissues are mounted on the microsphere coated glass slide with anti-fade reagents (Invitrogen) and covered with a cover plate. Fig. 4(a) shows the image of the brain tissue with a thickness of 100 μm. The microspheres in the tissue are shown in Fig. 4 (b).
Microspheres
(a)
(b)
Figure 4. Widefield microscopy images of mouse brain tissue (thickness=100 μm) with (a) 10x objective and (b) 60x water immersion objective. The focal plane is at the bottom of the tissue. The micro sphere is deposited on the glass slide.
3.2 Confocal imaging without wavefront compensation The spherical aberration induced by the cover glass was initially compensated by adjustment of a correction collar on the objective lens. The system aberration was further corrected by the DM by measuring the wavefront aberration from the microsphere at the bottom of the cover plate using the SHWS. The thickness of the brain tissue is 100µm. The confocal imaging system scans from the top surface to the bottom surface. The field of view is about 50µm. The frame rate is 30 frames/second. In order to improve the signal to noise ratio, 300 frames are averaged to generate the final image. Fig.5 (a) shows the confocal image at the top surface, where the cell body, dendrite and spine can be observed. Fig. 5(b) shows the image at the bottom surface. Because of the aberration induced by the tissue, it is very hard to find the dendrite and spine around the cell body.
(a)
(b)
Figure 5. Confocal imaging of mouse brain tissue without AO. The thickness of tissue is 100 μm. Figure (a) is the image on the top surface. Figure (b) is the confocal image at the bottom surface.
Proc. of SPIE Vol. 7931 79310L-5
3.3 Confocal imaging with wavefront compensation Brain tissues with thickness of 15 μm, 50 μm and 100 μm are evaluated in the system. A motorized Z stage under the sample focuses the HeNe laser on the microsphere at the bottom of the tissue. The wavefront error induced by the sample is measured by the SHWS. Twenty confocal images are collected by scanning along the Z axis with 3 μm range and 0.15 μm for each step. The final image is achieved by maximum intensity projection applied on these images. After turning on the DM, the wavefront error converges after around 10 iterations, which takes about 0.35 seconds. Then the confocal images are collected with the same setting as before correction.
4
4
x 10
4
x 10
2.5
2
2.5
1.5
2 1.5
1
1
Intensity
3 2
x 10
With AO Without AO
1.5 1 0.5
0.5 0.5 0
10
15
(b) 10000
8000
6000
12000
12000
10000
8000 6000
4000
4000 2000 2000 0
(d)
20
25
30
35
40
45
(c) 14000
10000
With AO Without AO
8000 6000 4000 2000
0
0 0
10
20
(e)
30
40
Pixel
50
60
(f) 8000
3000
With AO Without AO
2000
6000
1500 4000 1000 2000
500 0
0
(h)
Intensity
8000
2500
(g)
5
Pixel
Intensity
(a)
0 0
0
6000 4000 2000 0 0
10
20
30
Pixel
40
50
60
70
(i)
Figure 6. Confocal images with and without wavefront error correction and the intensity profile along the line indicated in the confocal images. Figure (a) and Figure (b) show the images before and after correction for brain tissue with 15 μm thickness. Figure (c) shows the intensity profile along the lines indicated in (a) and (b). Figure (d), (e) and (f) show the images before and after correction and intensity profile for brain tissues with 50 μm thickness. Figure (g), (h) and (i) show the results for brain tissue with 100 μm thickness.
The confocal images of the brain tissue with 15 μm thickness before correction and after correction are shown in Fig. 6 (a) and (b). The image of the dendrite and spines are much clearer after correction. The intensity profile along the line cross the dendrite and spine is shown in Fig. 6(c). Both the signal intensity and image contrast are improved. The signal intensity increases by 32%. The RMS wavefront error is 0.11λ (λ=633nm) before correction. After correction, RMS wavefront error was reduced to 0.01λ. Fig. 6 (d) and (e) show the confocal images of brain tissue with 50 μm thickness before and after correction. Because the system suffers more aberration, the image of the spines becomes dimmer. The intensity profile along the dash line and solid line is shown in Fig. 6(f). The signal intensity increases by 43%. The RMS wavefront error was reduced from 0.19λ to 0.03λ. For the brain tissue with 100 μm thickness, it is very hard to observe
Proc. of SPIE Vol. 7931 79310L-6
any feature as shown in Fig. 6(g). While after wavefront correction, the dendritic spine can be observed as shown in Fig. 6(h). The intensity profile is shown in Fig. 6(i). The signal intensity increases by 240%. The wavefront error before correction is shown in Fig. 7(a), which suffers a large amount of spherical aberration. The RMS error is 0.24 λ. After correction, the RMS decreased to 0.028 λ as shown in Fig. 7(b). The images of the microsphere before and after correction are shown in Figs. 7 (c) and (d). The relative Strehl ratio was measured using the method described in [15]. The improvement in the Strehl ratio was approximately 4.7x.
λ
λ
0.5
0.04
1
0.02
0.4
1
0.5
0.3 0.5
0
0.2
0
0.1
0
-0.02 -0.5
0 -0.5
-0.1 -0.2
-1 10
10 5
-0.3
10 5
-0.5
-0.06
5
-0.4
5 0
-0.04
-1 10
0
0
0
( ) (a)
((b))
5 10
200
200
150
150
100
100
50
50
0
0
15 20 25 30 35 40 45 50 55
10
20
30
40
50
((c) )
(d) (d)
Figure 7. Wavefront measurements from a fluorescent microsphere for 100 μm thick brain tissue. Wavefront error before correction (a) and after correction (b). The root mean square (RMS) errors for (a) and (b) are 0.24λ and 0.028λ (λ=633 nm), respectively. Images of the microsphere before correction (c) and after correction (d).
4. CONCLUSION Optical aberrations due to the inhomogeneous refractive index of tissue degrade the resolution and brightness of images in deep tissue imaging. We have demonstrated that with the use of direct wavefront measurement by a SHWS and a DM, the AOCFM can effectively measure and correct the aberration at a high speed, which is important for live in-vivo imaging. The experiment for the fixed brain tissue sample shows a great improvement in the final after wavefront error correction. The microsphere injection in the brain tissue and live in-vivo imaging will be investigated in the future.
5. ACKNOWLEDGMENTS This material is based upon work supported by the National Science Foundation under Grant No. 0852742. We would like to thank Claire Max at the Center for Adaptive Optics, UCSC for useful conversations regarding this project, Donald Gavel and Daren Dillon at the Laboratory for Adaptive Optics, UCSC, for access to their facility, Yu-Chen Hwang from UCSC Life Sciences Microscopy Center for technical support.
Proc. of SPIE Vol. 7931 79310L-7
REFERENCES [1] Okubo, Y., Kakizawa, S., Hirose, K. and Iino, M., “Visualization of IP(3) dynamics reveals a novel AMPA receptor-triggered IP(3) production pathway mediated by voltage-dependent Ca(2+)influx in Purkinje cells”, Neuron 32, 113–122 (2001). [2] Guerrero, G., Siegel, M. S., Roska, B., Loots, E., Isacoff, E.Y., “Tuning FlaSh: redesign of the dynamics, voltage range, and color of the genetically encoded optical sensor of membrane potential”, Biophys J. 83, 3607–3618 (2002). [3] Grutzendler, J., Kasthuri, N. and Gan, W. B., “Long-term dendritic spine stability in the adult cortex”, Nature 420, 812–816 (2002). [4] Chiu, C. S., Jensen, K., Sokolova, I., Wang, D., Li, M., Deshpande, P., Davidson, N., Mody, I., Quick, M. W., Quake, S. R. and Lester, H. A., “Number, density, and surface/cytoplasmic distribution of GABA transporters at presynaptic structures of knock-in mice carrying GABA transporter subtype 1-green fluorescent protein fusions”, J. Neurosci 22, 10251–10266 (2002). [5] Beuthan, J., Minet, O., Helfmann, J., Herrig, M., and Muller, G., “The spatial variation of the refractive index in biological cells”, Phys. Med. Biol. 41, 369–382 (1996). [6] Xu, H., Pan, F., Yang, G., and Gan, W., Choice of cranial window type for in vivo imaging affects dendritic spine turnover in the cortex, Nature Neuroscience 10, 549 - 551 (2007) [7] Pawley, J. B., Handbook of Biological Confocal Microscopy, Springer Science (2006). [8] Booth, M.J., “Adaptive optics in microscopy”, Phil. Trans. R. Soc. A–Math. Phys. Eng. Sci. 365, 2829–2843 (2007). [9] R. K. Tyson, Principles of adaptive optics, Academic ( 1991). [10] Debarre, D., Botcherby, E.J., Watanabe, T., Srinivas, S., Booth, M.J., and Wilson, T., “Image-based adaptive optics for two-photon microscopy”, Opt. Lett. 34, 2495–2497 (2009). [11] Marsh, P., Burns, D., and Girkin, J., “Practical implementation of adaptive optics in multiphoton microscopy”, Opt. Express 11, 1123–1130 (2003). [12] Albert O., Sherman, L., Mourou, G., Norris, T.B., and Vdovin, G., “Smart microscope: an adaptive optics learning system for aberration correction in multiphoton confocal microscopy”, Opt. Lett. 25, 52-54 (2000). [13] Ji, N., Milkie, D.E., and Betzig, E., “Adaptive optics via pupil segmentation for high-resolution imaging in biological tissues”, Nat. Methods 7, 141–147 (2010). [14] Rueckel, M., Mack-Bucher, J.A., and Denk, W., “Adaptive wavefront correction in two-photon microscopy using coherence-gated wavefront sensing,” Proc. Natl. Acad. Sci. U.S.A. 103, 17137–17142 (2006). [15] Azucena, O., Crest, J., Cao, J., Sullivan, W., Kner, P., Gavel, D., Dillon, D., Olivier, S., and Kubby, J., “Wavefront aberration measurements and corrections through thick tissue using fluorescent microsphere reference beacons”, Opt. Express 18, 17521-17532 (2010). [16] Poyneer, A. L, Gavel, D. T., and Brase, J.M., “Fast wave-front reconstruction in large adaptive optics systems with use of the Fourier transform,” J. Opt. Soc. Am. A 19, 2100–2111 (2003). [17] Thomas, S., Fusco, T., Tokovinin, A., Nicolle, M., Michau, V., and Rousset, G., “Comparison of centroid computation algorithms in a Shack-Hartmann sensor,” Mon. Not. R. Astron. Soc. 371, 323–336 (2006). [18] Porter, J., Queener, H., Lin, J., Thorn, K., and Awwal, A.A.S, “Adaptive Optics for Vision Science: Principles, Practices, Design and Applications”, Wiley (2006).
Proc. of SPIE Vol. 7931 79310L-8
1. INTRODUCTION Fluorescence microscopy has been widely used for the study of the neuron, such as monitoring biochemical signaling pathways [1], visualization of neuronal dynamics [2], determinations of cellular morphologies [3] and mapping protein distribution in cells [4]. For imaging of the long dendrite and tiny dendritic spines of the neuron, high-resolution fluorescence imaging deep within brain tissue is required. Although some objectives are equipped with a correction collar for compensation of the spherical aberration from the cover slide, the high numerical objectives that are used still suffer aberration from the inhomogeneous refractive index of the tissue and the refractive index mismatch between the sample and mounting medium [5]. For in-vivo imaging, the non-uniform skull thickness may also cause significant spherical aberrations [6]. Compared with the wide-field microscope, the confocal microscope suffers from more aberration because both the excitation and emission light are distorted when they go though the sample [7]. The distortion in excitation light will decrease the intensity of the fluorescent emission light. Using a higher power laser to compensate this effect will cause phototoxicity and photobleaching [8]. Both of these distortions also reduce the resolution of the system. Moreover, the distortion increases with the imaging depth. A similar issue can be found in astronomy where the aberration induced by atmospheric turbulence degrades the final image from the telescope [9]. In order to compensate this dynamic aberration, the wavefront distortion is measured directly by a wavefront sensor. An active optical component, such as deformable mirror (DM) or spatial light modulator (SLM), corrects the aberration in a feed back loop [9]. However, wavefront sensors often need a reference point source, such as laser guide star, which is difficult to realize in biological specimens. Because of this, most of the existing adaptive optics (AO) confocal microscopes developed so far are based on indirect methods of wavefront measurement which depends on processing of the final image [10,11,12,13]. In order to find the wavefront parameters, numerous iterations are required, which will cause photobleaching and limits the bandwidth for live imaging. Backscattered light has been used for direct wavefront sensing [14]. However it is highly dependent on the backscattering efficiency of the tissue. The complex procedure used for wavefront sensing makes it computationally intensive. *[email protected]; phone (831) 295-3957 MEMS Adaptive Optics V, edited by Scot S. Olivier, Thomas G. Bifano, Joel A. Kubby, Proc. of SPIE Vol. 7931, 79310L · © 2011 SPIE · CCC code: 0277-786X/11/$18 · doi: 10.1117/12.876524
Proc. of SPIE Vol. 7931 79310L-1
In this paper, we describe an adaptive optics confocal fluorescence microscope (AOCFM) which can correct aberrations based on direct wavefront measurements using a point source reference beacon and a Shack-Hartmann Wavefront Sensor (SHWS). In order to measure the wavefront directly, fluorescent microspheres with a diameter of 1µm can be injected into the mouse brain tissue. These microspheres were used as reference beacons for the adaptive AO system, similar to the artificial guide stars used in astronomical AO systems [15]. The microsphere is excited by a HeNe laser (633nm). The emission from the microsphere goes through a filter, which blocks the emission from the tissue, and is finally collected by a SHWS comprised of a lenslet array focused on a cooled CCD camera. The control signal for the deformable mirror is calculated based on the estimated wavefront error from the SHWS. In the experiment, mouse brain tissues with different thicknesses are placed on top of the microsphere reference beacon to test the proposed AO system. The neurons were marked by the transgenic expression of yellow fluorescent protein (YFP).
2. METHOD 2.1 Optical System Layout Fig.1. Shows the layout of the AOCFM used for brain tissue imaging. The system was designed and optimized using the optical design software (CODE V). A 60X water immersion objective with a numerical aperture of 1.2 was used (Olympus Microscope, Center Valley, PA) for imaging of the dendrites and spines in brain tissue. The optical system includes three telescope relay sub-systems. Lenses L1 and L2 image the exit pupil of objective onto the Y scanner. Lenses L3 and L4 relay the X scanner conjugate to the Y scanner. This design minimizes the movement of scanning beam at the exit pupil of objective and the emission light at the DM, which is important for accurate wavefrront measurement and correction. Lenses L2 and L3 also serve as scanning lenses. The current design is optimized for an optical scanning angle of 4.4 degree, which provides a field of view of 128 µm on the sample with a 60x objective. By changing the control signal of the scanners, the field of view can be easily adjusted. Lens L5 and L6 image the pupil of the X scanner onto the DM. Lenses L7 and L8 image the pupil of DM onto the wavefront sensor. The diameter of the effective aperture on the DM used in this design is 4 mm. The exit pupil of the objective is 7.2 mm. To match the aperture of the DM with the objective, the telescope L1 and L2 demagnifies the pupil from 7.2 mm to 4mm.
L10 L9
L8
L7 L6 L5
L4
L3
L2 L1
L: Lens
Figure 1. System Layout of the Adaptive Optics Confocal Microscope. A HeNe laser emits light at 633nm for excitation of the fluorescent reference beacons. A solid state laser emits light at 515 nm that excites yellow fluorescent protein (YFP) bred into the sample. F2 and F4 are excitation filters. Light emitted from the reference beacon is passed through filter F1 to the wavefront sensor. The dichroic beamsplitters DB2 separate the light from reference beacon and the sample. The dichroic beamspliter DB1 and DB3 are used to separate the excitation light and its associated back reflection. The fluorescent light emitted by the YFP is filtered by F3 and detected by the Photo-Multiplier Tube (PMT). The wavefront aberration is corrected by the Deformable Mirror (DM).
Proc. of SPIE Vol. 7931 79310L-2
Two laser channels are included in the system, a solid state laser (λ=515nm) and a helium neon laser (λ=633nm) used for confocal fluorescence imaging and wavefront sensing, respectively. The two channels share the same lightpath through the DM, relay lens, scanner and scanning lens and feed into an Olympus IX71 inverted microscope (Olympus Microscope, Center Valley, PA) from its side optical port. The focused beam is scanned on the sample in a raster pattern with a resonant scanner (SC-30, 16 kHz, 5°, Electro-Optics Products Corp) and a vertical scanner (GVS001, Thorlabs). The control signals for the scanners are generated by a data acquisition board (PCIe-6363, National Instruments Corporation). The emission light from both the microsphere and the sample are collected by an objective lens (Olympus 60x, 1.2 NA water immersion) and separated by a dichroic mirror (DB2). The light from the sample was focused on a pin hole by an achromatic lens L10. A GaAs photomultiplier tube (PMT) (H422-50, Hamamatsu) was used as a photon detector. The signal from the PMT is then fed to a frame-grabbing board (Helios eA/XA, Matrox Imaging). With the vertical and horizontal synchronized signal from the DAQ, the frame-grabbing board generates the raw image of 512x512 pixels at 30 frames per second. When the signal levels are low, more frames can be averaged to increase the signal to noise ratio. The experimental system was setup on an optical table with vibration isolation as shown in Fig. 2. The beam from the scanning system was fed into Olympus IX71 inverted microscope by a periscope. The size of the system without the microscope frame is about 48 inch ×28 inch.
Science Camera
OLYMPUS X61 Frame
PMT HeNe Laser
Solid State Laser
DM
SH Wavefront sensor
X Scanner Y Scanner Figure 2. Experimental setup of Adaptive Optics Confocal Microscope. The system was mounted on an optical table with vibration isolation.
2.2 Wavefront Sensing To correct the aberration induced by the specimen, accurate measurement of the wavefront becomes an important factor. SHWS has been widely used in astronomy and vision science, which consists of a lenslet array and a camera at its focal plane. By measuring the displacement of spots imaged by the lenslet array from their reference positions, the wavefront error can be estimated [16]. In the current design, the light from the microsphere is fed into the SHWS with 44×44 element lenslet array (AOA Inc., Cambridge, MA) and an electron multiplying CCD camera (Photometrics). The pitch of the lenslet array is 400 µm, which is equal to the pitch of the actuators in the DM. The SHWS requires a reference source to produce a spot image on the sensor. In astronomy, a real star or a laser guid star are often used as the reference source [9]. However, it is difficult to find one in the biological sample. In order to measure the wavefront directly, fluorescent microspheres with a diameter of 1µm can be imbedded in the sample as
Proc. of SPIE Vol. 7931 79310L-3
reference source (FluoSpheres crimson fluorescent microspheres, Invitrogen, Carlsbad, CA) [15]. In order to reduce the overlap between the biology sample and fluorescent microspheres on both the excitation and the emission side, the microspheres and dichroic filter are selected based on the spectrum of the fluorescent protein in the biology sample. In this study, the neuron in the brain tissue is YFP labeled. Its peak absorption and emission wavelength are 514 nm and 527 nm. A solid state laser ((λ=515 nm) was used as the excitation source. The spectrum profile is show in Fig. 3. A single-band laser filter set for YFP (LF514-A, Semrock, Inc.) is used in the system. The transmittance profile of the emission filter in the filter set is shown as a purple line in Fig. 3. The absorption and emission spectra profile of the microsphere is also shown in Fig. 3. The overlap between the emissions from YFP and microsphere is only 3% of the total emission power. The peak absorption and emission wavelength of the microsphere are 625 nm and 645 nm. A HeNe laser was used as the excitation source. An ultra-steep long-pass edge filter (LP02-633RU, Semrock, Inc.) is used as the emission filter which was mounted on the wavefront sensor. 83% of emission light from the microsphere passes though the filter. Only 2.7% of emission light from YFP goes through the filter. The emission filter for the YFP can block the light from the microsphere. By demixing the light path between confocal imaging and wavefront sensing channels, this design reduces the noise in wavefront measurement and blocks the light from the microsphere in the final confocal image. Before wavefront measurement, a SH reference image was captured when a collimated laser beam was added to the optical path after the DM. The position of the spots in the image from the wavefront sensor was detected using a crosscorrelation centroiding algorithm [17]. During the wavefront measurement, the wavefront slopes can be obtained from the displacement of spots in the image. Then a Fast Fourier Transform (FFT) reconstruction algorithm was implemented to obtain the real wavefront [16]. It can measure the wavefront induced from the sample, mounting medium, cover slide and misaligned optical components. Excitation source for microsphere (633nm)
Excitation source for YFP (515nm) 100 90
70
90 80 70
60
60
50
50
40
40
30
30
20
20
10
10
0 300
350
400
450
500
550
600
650
700
750
800
Relative Intensity (%)
80
Transmittance (%)
100
YFP Emission YFP Absorption Microsphere Emission Microsphere Absorption Emission Filter for Microsphere Emission Filter for YFP
0
Figure 3. The absorption/emission spectral profile and transmittance profile of the emission filters for YFP and fluorescent microsphere. The wavelengths of the excitation source for YFP and the microsphere are 515 nm and 633 nm, respectively.
2.3 Wavefront Compensation In order to correct the wavefront, a DM (Boston Micromachines) with 140 actuators and 3.5 µm of stroke is placed in the optical path, where it is conjugate to the exit pupil of the objective, the X and Y scanners and the wavefront sensor. Because the excitation and emission light shares the same path, the DM can correct the aberration on both paths. The effective aperture size of the DM used in this design is 4 mm, which is the same as the wavefront sensor. With the correction information from the wavefront sensor, a direct slope algorithm was applied to control the DM in a closed loop [18]. The influence matrix of the DM was calibrated first with the microsphere under the cover slide outside of the sample. For the real wavefront compensation, the slopes of the wavefront were measured from the displacement of the spots on the wavefront sensor. The control signal for the DM is then calculated by multiplying the pseudo inverse of the influence matrix of the DM and the slope measurement with a proportional gain of 0.4.
Proc. of SPIE Vol. 7931 79310L-4
3. EXPERIMENTAL RESULT 3.1 Sample Preparation To investigate the feasibility of the proposed system, a fixed brain slice from YFP-H line transgenic mouse was prepared. YFP is labeled on the cell body and protrusions of neurons. One micron diameter crimson fluorescent microspheres (Invitrogen, Carlsbad, CA) were deposited onto a glass slide and a cover plate for use as laser guide stars. Sample brain coronal sections of different thicknesses (15 μm, 50 μm and 100 μm) were cut with a microtome. The tissues are mounted on the microsphere coated glass slide with anti-fade reagents (Invitrogen) and covered with a cover plate. Fig. 4(a) shows the image of the brain tissue with a thickness of 100 μm. The microspheres in the tissue are shown in Fig. 4 (b).
Microspheres
(a)
(b)
Figure 4. Widefield microscopy images of mouse brain tissue (thickness=100 μm) with (a) 10x objective and (b) 60x water immersion objective. The focal plane is at the bottom of the tissue. The micro sphere is deposited on the glass slide.
3.2 Confocal imaging without wavefront compensation The spherical aberration induced by the cover glass was initially compensated by adjustment of a correction collar on the objective lens. The system aberration was further corrected by the DM by measuring the wavefront aberration from the microsphere at the bottom of the cover plate using the SHWS. The thickness of the brain tissue is 100µm. The confocal imaging system scans from the top surface to the bottom surface. The field of view is about 50µm. The frame rate is 30 frames/second. In order to improve the signal to noise ratio, 300 frames are averaged to generate the final image. Fig.5 (a) shows the confocal image at the top surface, where the cell body, dendrite and spine can be observed. Fig. 5(b) shows the image at the bottom surface. Because of the aberration induced by the tissue, it is very hard to find the dendrite and spine around the cell body.
(a)
(b)
Figure 5. Confocal imaging of mouse brain tissue without AO. The thickness of tissue is 100 μm. Figure (a) is the image on the top surface. Figure (b) is the confocal image at the bottom surface.
Proc. of SPIE Vol. 7931 79310L-5
3.3 Confocal imaging with wavefront compensation Brain tissues with thickness of 15 μm, 50 μm and 100 μm are evaluated in the system. A motorized Z stage under the sample focuses the HeNe laser on the microsphere at the bottom of the tissue. The wavefront error induced by the sample is measured by the SHWS. Twenty confocal images are collected by scanning along the Z axis with 3 μm range and 0.15 μm for each step. The final image is achieved by maximum intensity projection applied on these images. After turning on the DM, the wavefront error converges after around 10 iterations, which takes about 0.35 seconds. Then the confocal images are collected with the same setting as before correction.
4
4
x 10
4
x 10
2.5
2
2.5
1.5
2 1.5
1
1
Intensity
3 2
x 10
With AO Without AO
1.5 1 0.5
0.5 0.5 0
10
15
(b) 10000
8000
6000
12000
12000
10000
8000 6000
4000
4000 2000 2000 0
(d)
20
25
30
35
40
45
(c) 14000
10000
With AO Without AO
8000 6000 4000 2000
0
0 0
10
20
(e)
30
40
Pixel
50
60
(f) 8000
3000
With AO Without AO
2000
6000
1500 4000 1000 2000
500 0
0
(h)
Intensity
8000
2500
(g)
5
Pixel
Intensity
(a)
0 0
0
6000 4000 2000 0 0
10
20
30
Pixel
40
50
60
70
(i)
Figure 6. Confocal images with and without wavefront error correction and the intensity profile along the line indicated in the confocal images. Figure (a) and Figure (b) show the images before and after correction for brain tissue with 15 μm thickness. Figure (c) shows the intensity profile along the lines indicated in (a) and (b). Figure (d), (e) and (f) show the images before and after correction and intensity profile for brain tissues with 50 μm thickness. Figure (g), (h) and (i) show the results for brain tissue with 100 μm thickness.
The confocal images of the brain tissue with 15 μm thickness before correction and after correction are shown in Fig. 6 (a) and (b). The image of the dendrite and spines are much clearer after correction. The intensity profile along the line cross the dendrite and spine is shown in Fig. 6(c). Both the signal intensity and image contrast are improved. The signal intensity increases by 32%. The RMS wavefront error is 0.11λ (λ=633nm) before correction. After correction, RMS wavefront error was reduced to 0.01λ. Fig. 6 (d) and (e) show the confocal images of brain tissue with 50 μm thickness before and after correction. Because the system suffers more aberration, the image of the spines becomes dimmer. The intensity profile along the dash line and solid line is shown in Fig. 6(f). The signal intensity increases by 43%. The RMS wavefront error was reduced from 0.19λ to 0.03λ. For the brain tissue with 100 μm thickness, it is very hard to observe
Proc. of SPIE Vol. 7931 79310L-6
any feature as shown in Fig. 6(g). While after wavefront correction, the dendritic spine can be observed as shown in Fig. 6(h). The intensity profile is shown in Fig. 6(i). The signal intensity increases by 240%. The wavefront error before correction is shown in Fig. 7(a), which suffers a large amount of spherical aberration. The RMS error is 0.24 λ. After correction, the RMS decreased to 0.028 λ as shown in Fig. 7(b). The images of the microsphere before and after correction are shown in Figs. 7 (c) and (d). The relative Strehl ratio was measured using the method described in [15]. The improvement in the Strehl ratio was approximately 4.7x.
λ
λ
0.5
0.04
1
0.02
0.4
1
0.5
0.3 0.5
0
0.2
0
0.1
0
-0.02 -0.5
0 -0.5
-0.1 -0.2
-1 10
10 5
-0.3
10 5
-0.5
-0.06
5
-0.4
5 0
-0.04
-1 10
0
0
0
( ) (a)
((b))
5 10
200
200
150
150
100
100
50
50
0
0
15 20 25 30 35 40 45 50 55
10
20
30
40
50
((c) )
(d) (d)
Figure 7. Wavefront measurements from a fluorescent microsphere for 100 μm thick brain tissue. Wavefront error before correction (a) and after correction (b). The root mean square (RMS) errors for (a) and (b) are 0.24λ and 0.028λ (λ=633 nm), respectively. Images of the microsphere before correction (c) and after correction (d).
4. CONCLUSION Optical aberrations due to the inhomogeneous refractive index of tissue degrade the resolution and brightness of images in deep tissue imaging. We have demonstrated that with the use of direct wavefront measurement by a SHWS and a DM, the AOCFM can effectively measure and correct the aberration at a high speed, which is important for live in-vivo imaging. The experiment for the fixed brain tissue sample shows a great improvement in the final after wavefront error correction. The microsphere injection in the brain tissue and live in-vivo imaging will be investigated in the future.
5. ACKNOWLEDGMENTS This material is based upon work supported by the National Science Foundation under Grant No. 0852742. We would like to thank Claire Max at the Center for Adaptive Optics, UCSC for useful conversations regarding this project, Donald Gavel and Daren Dillon at the Laboratory for Adaptive Optics, UCSC, for access to their facility, Yu-Chen Hwang from UCSC Life Sciences Microscopy Center for technical support.
Proc. of SPIE Vol. 7931 79310L-7
REFERENCES [1] Okubo, Y., Kakizawa, S., Hirose, K. and Iino, M., “Visualization of IP(3) dynamics reveals a novel AMPA receptor-triggered IP(3) production pathway mediated by voltage-dependent Ca(2+)influx in Purkinje cells”, Neuron 32, 113–122 (2001). [2] Guerrero, G., Siegel, M. S., Roska, B., Loots, E., Isacoff, E.Y., “Tuning FlaSh: redesign of the dynamics, voltage range, and color of the genetically encoded optical sensor of membrane potential”, Biophys J. 83, 3607–3618 (2002). [3] Grutzendler, J., Kasthuri, N. and Gan, W. B., “Long-term dendritic spine stability in the adult cortex”, Nature 420, 812–816 (2002). [4] Chiu, C. S., Jensen, K., Sokolova, I., Wang, D., Li, M., Deshpande, P., Davidson, N., Mody, I., Quick, M. W., Quake, S. R. and Lester, H. A., “Number, density, and surface/cytoplasmic distribution of GABA transporters at presynaptic structures of knock-in mice carrying GABA transporter subtype 1-green fluorescent protein fusions”, J. Neurosci 22, 10251–10266 (2002). [5] Beuthan, J., Minet, O., Helfmann, J., Herrig, M., and Muller, G., “The spatial variation of the refractive index in biological cells”, Phys. Med. Biol. 41, 369–382 (1996). [6] Xu, H., Pan, F., Yang, G., and Gan, W., Choice of cranial window type for in vivo imaging affects dendritic spine turnover in the cortex, Nature Neuroscience 10, 549 - 551 (2007) [7] Pawley, J. B., Handbook of Biological Confocal Microscopy, Springer Science (2006). [8] Booth, M.J., “Adaptive optics in microscopy”, Phil. Trans. R. Soc. A–Math. Phys. Eng. Sci. 365, 2829–2843 (2007). [9] R. K. Tyson, Principles of adaptive optics, Academic ( 1991). [10] Debarre, D., Botcherby, E.J., Watanabe, T., Srinivas, S., Booth, M.J., and Wilson, T., “Image-based adaptive optics for two-photon microscopy”, Opt. Lett. 34, 2495–2497 (2009). [11] Marsh, P., Burns, D., and Girkin, J., “Practical implementation of adaptive optics in multiphoton microscopy”, Opt. Express 11, 1123–1130 (2003). [12] Albert O., Sherman, L., Mourou, G., Norris, T.B., and Vdovin, G., “Smart microscope: an adaptive optics learning system for aberration correction in multiphoton confocal microscopy”, Opt. Lett. 25, 52-54 (2000). [13] Ji, N., Milkie, D.E., and Betzig, E., “Adaptive optics via pupil segmentation for high-resolution imaging in biological tissues”, Nat. Methods 7, 141–147 (2010). [14] Rueckel, M., Mack-Bucher, J.A., and Denk, W., “Adaptive wavefront correction in two-photon microscopy using coherence-gated wavefront sensing,” Proc. Natl. Acad. Sci. U.S.A. 103, 17137–17142 (2006). [15] Azucena, O., Crest, J., Cao, J., Sullivan, W., Kner, P., Gavel, D., Dillon, D., Olivier, S., and Kubby, J., “Wavefront aberration measurements and corrections through thick tissue using fluorescent microsphere reference beacons”, Opt. Express 18, 17521-17532 (2010). [16] Poyneer, A. L, Gavel, D. T., and Brase, J.M., “Fast wave-front reconstruction in large adaptive optics systems with use of the Fourier transform,” J. Opt. Soc. Am. A 19, 2100–2111 (2003). [17] Thomas, S., Fusco, T., Tokovinin, A., Nicolle, M., Michau, V., and Rousset, G., “Comparison of centroid computation algorithms in a Shack-Hartmann sensor,” Mon. Not. R. Astron. Soc. 371, 323–336 (2006). [18] Porter, J., Queener, H., Lin, J., Thorn, K., and Awwal, A.A.S, “Adaptive Optics for Vision Science: Principles, Practices, Design and Applications”, Wiley (2006).
Proc. of SPIE Vol. 7931 79310L-8
