|Home | About | Journals | Submit | Contact Us | Français|
Spatial Spectral Holographic imaging system (S2-VHIS) is a promising alternative to confocal microscopy due to its capabilities to simultaneously image several sample depths with high resolution. However, the field of view of previously presented S2-VHIS prototypes has been restricted to less than 200μm. This paper presents experimental results of an improved S2-VHIS design which have a field of view of ~1mm while maintaining high resolution and dynamic range.
Spectroscopic and microscopic imaging instruments are essential tools in the biological and medical sciences. Instruments that combine both techniques into the same instrument can further enhance the understanding of biological processes. Sophisticated systems such as combined optical coherence tomography/laser induced fluorescence , multispectral confocal microendoscopy , and computed tomographic imaging spectroscopy  provide the capability for collecting both spatial and spectral data about an object simultaneously. However, each requires the use of mechanical scanning to build up an entire 4-D (x,y,z,λ) data set that represents a physical object.
A spectral-spatial volume holographic imaging system (S2-VHIS) is an optical device that minimizes the need for mechanical scanning. Therefore, for an given volume, images can be obtained faster using S2-VHIS than using confocal microscopes or optical coherent tomography systems (OCT). The limitations of S2-VHIS and confocal microscopes are the depth penetration which limited to < 200 microns compared with <1mm of OCT systems. In terms of lateral resolution, S2-VHIS can theoretically achieve similar resolution than confocal microscopes. In practice, this requires a lot of improvements in the spectrum-angular selectivity of the hologram. S2-VHIS depth resolution is comparable currently comparable with OCT systems.
The key to this device is a volume holographic element that converts a four-dimensional (4-D) distribution of information to a two-dimensional distribution. The 4-D distribution being converted is, in general, a collection of point scattering elements with dimensions (x,y,z) and band-limited emission spectra with mean wavelength λ0 and bandwidth Δλ.
The functionalities of S2-VHIS system has been previously described [4-7]. Conceptual designs have been implemented in LiNbO3 [5,6] and more recently in doped polymer  achieving excellent optical sectioning capabilities in a field of view <200 μm.
In this paper, we present an improved S2-VHIS design that enables the observation of a field several times larger, while maintaining micron-scale resolution.
The remainder of the paper consists in five sections. In Section 2 we summarize the operation of S2-VHIS. Section 3 details the system requirements to improve the field of view. The design of the components is described in Section 4. Experimental results are shown in Section 5. Section 6 summarizes our results.
As shown in Figure 1, S2-VHIS consists of a 4-f configuration of lenses with a holographic element placed in the Fourier plane. The holographic element is composed of angle-multiplexed planar and spherical wave gratings. The nominal gratings for this device are optimized for infinite-conjugates, where a spherical (or plane) wave incident upon the hologram generates a diffracted plane wave. The holographic element is considered a thick phase grating, and can be analyzed exactly using coupled wave theory [7-9], or Fourier Optics in the weak diffraction approximation .
The diffraction efficiency of the hologram is very sensitive to the wavelength and shape of the incident wavefront. Due to this feature, the hologram acts as a wavefront filter. The action of the filter is to select only those wavefronts that originate at a particular depth from within the scattering object which are within the designed spectral band. The optical section width (axial or z resolution) is determined by how well the hologram is able to reject wavefronts generated by out of plane point sources. An analytical solution quantifying the optical sectioning performance of this system has been derived for simple planar grating structures [5,6]. More complicated structures require the use of numerical analysis [9,10]. For the purposes of designing the system architecture, all that is necessary for simulation of the hologram is the wavelength-angle relationship defined by the Bragg condition of a planar grating . This condition gives the correct angle of incidence (θ) of the on-axis beam for a given wavelength. The optical axis of the system is rotated bent by twice the angle of incidence, giving the equation Φ(λ) = 2θ(λ). Thus, the wavelength and x axes are non-orthogonal, but they may be decoupled by using multiplexed gratings with matched response to specific wavelengths/wavefronts. Dispersion in the hologram ultimately limits the spatial resolution along the x axis, as well as the spectral resolution. The lack of grating periodicity along the y axis results in a degeneracy that provides the y axis field of view. In the diffraction limit, the resolution along this axis is limited by the numerical aperture of the objective lens.
Multiplexing many gratings into the same volume allows for the observation of multiple planes in the object simultaneously . Given properly designed multiplexed gratings, it is possible to map any point (x, y, z) in object space to a point (x, y) in image space. The holographic sectioning capability of the S2-VHIS bears some resemblance to the action of a slit in confocal microscopy  with two differences: (1) the multiplexing property of volume holograms allows sections from multiple “slits” to be viewed simultaneously, and (2) the dispersion property of volume gratings allows the instrument to also have spectral sectioning capability. Also, there are no special illumination conditions for this system other than that the object must be incoherently radiating light with some appreciable bandwidth. This means that reflection and transmission with spatially incoherent illumination, and fluorescence modes of operation are valid for this instrument.
We designed an improved S2-VHIS that enables the observation of a field several times larger than previously realized, while maintaining micron-scale resolution. According to , the system performs well when using objective and collector lenses corrected for infinite conjugates. This is because near-collimated light reduces aberrations in the diffracted beam produced by the hologram. As this system is intended for use in microscopy, the system was designed to be telecentric in object space to maintain constant lateral magnification with axial location in the sample. For this system to be telecentric in object space and unvignetted over an appreciable field of view, the hologram must be the limiting aperture in the system, otherwise known as the aperture stop.
The previous system, as illustrated in Figure 1, used commercially available long working distance microscope objectives and suffered from vignetting. A common feature of lenses of this type is that that the aperture stop is at an inaccessible location inside the objective housing. Therefore, it is not feasible to place the hologram with the pupil without auxiliary optics. This results in vignetting, as shown in Figure 2(a). The remedy to this problem is to insert a telecentric relay into the system such that the aperture stop of the objective lens is imaged onto the hologram, as shown in Figure 2(b).
The resolutions along the lateral and the axial dimensions are primarily determined by the numerical aperture of the objective lens and the thickness of the holographic grating, respectively. It was experimentally determined that an objective lens numerical aperture of 0.55 (fo = 3.6 mm) and a grating thickness of 1.8 mm leads to an axial resolution of 22 μm and lateral resolution of 4.5 μm for a wavelength of 633 nm. Increasing the objective NA or hologram thickness can reduce the section thickness further [5,6]. The collector lens’ focal length (fc) is chosen based upon the focal length of the objective lens and the desired lateral magnification |fc/fo|. A Lumenera 6 Mpixel CCD array with a pixel pitch of 3.5 μm is used for imaging in reflectance mode, and a highly sensitive Andor iXon array with a pixel pitch of 16 μm is used for fluorescence mode imaging. In combination with these cameras, collector focal lengths of 10 mm (Lumenera) and 20 mm (Andor) are used to resolve lateral features smaller than 15 μm. Both objective and collector lenses are achromatic, because the system functions over a 100 nm band in the visible wavelength range of 486 - 656 nm. The lenses used are commercially available microscope objectives with long working distances. The objective lens is an Olympus ULWDMSPlan50X (NA = 0.55, f = 3.6 mm), and the collector lenses are Mitutuyo M Plan APO 20x (NA = 0.42, f = 10.0 mm) and M Plan APO 10x (NA = 0.28, f = 20.0 mm).
The form of the telecentric relay system can be approached from 1st order paraxial optics, and basic 3rd order aberration analysis. A focal length of 33mm is used for each element, as this focal length allows sufficient working distance at both ends of the relay while using standard 1” diameter optics. The wavefront error tolerance for the relay is derived from the optical section thickness Δz. The maximum allowed wavefront distortion that can be tolerated by the hologram may be approximated by defocus, and is given by . For a 0.55 NA objective at a wavelength of 633 nm, this corresponds to a wavefront change of approximately 2.6λ for Δz~ 22 μm. Therefore, an OPD change of 2.6λ between the on-axis and full-field object points is all that is allowable in order to achieve high contrast optical slicing over the field of view.
Since the design is symmetrical, odd aberrations are cancelled and only even aberrations need be considered. Of the even aberrations, field curvature is of less importance since modest curvature in the object plane is acceptable. The aberrations of main concern are therefore spherical aberration, astigmatism, and axial chromatic aberration. Achromatic doublets were chosen for our design in order to correct for spherical and chromatic aberrations. The relay as designed uses a combination of catalog achromatic doublets, and can be thought of as two back-to-back Ploessl-type eyepieces , where each Ploessl system consists of two achromats which are facing each other. Aberrations are sufficiently optimized to provide a MTF contrast greater than 0.6 for a 15 μm feature over a 1.2 mm field of view, although the full un-vignetted field of view is ~2 mm.
A multiplexed hologram was formed by superimposing interferometric exposures in the same volume of phenathrenequinone (PQ) doped poly(methyl methacrylate) (PMMA) photopolymer recording material. The material was recorded using an Argon Ion Laser operating at a wavelength of 0.488 μm. A full-width half maximum (FWHM) angular bandwidth of ~0.03° was obtained for each grating. The FWHM spectral resolution was 0.2 nm.
With the improved optical design, a well-corrected y field of view of 1.2 mm has been achieved. Earlier experiments had established that without the addition of the relay systems, only about 125 μm of the sample had been visible. The field of view in the x dimension can be varied depending on the hologram design and the spectral bandwidth of the emitted light. In one realized system it is 220 μm with two multiplexed gratings and reflected light from an LED source with center wavelength of 633 nm and bandwidth of 25 nm (FWHM). The LED used was a Cree XLamp XR7090RED. The Air Force Resolution Chart image in Figure 5a, taken in transmission mode, shows that lateral features as small as 4.5 μm are well resolved by this system. The resolution along the x axis is slightly worse than that along the y axis due to dispersion in the hologram. A reflectance mode image taken using a human ovary sample shows two depth sections captured simultaneously, where the left-hand image is taken just below the surface of the ovary, and the right-hand image is designed to be approximately 70 μm deeper in the tissue. Another realization of the instrument uses fluorescent stains as the dominant contributor to the object's spectral signal (Figure 5c). A sample of mouse colon was stained with acridine orange, then illuminated with a 355 nm wavelength laser. The emission spectrum of the fluorophore had an emission bandwidth of approximately 100 nm at a center wavelength of ~550 nm. This wide bandwidth increases the x field of view to 0.5 mm; thus the hologram was designed for greater angular separation of depth sections.
The depth sectioning capabilities of the improved S2-VHIS design were also evaluated. A point source was generated using a collimated 633 nm HeNe laser and a microscope objective placed on a motorized translation stage. Results shown in Figure 6, indicates an FWHM axial resolution of 22.3 μm. Crosstalk between depths separated >70 μm are < 10%. This value can be further reduced by applying image processing techniques.
In this paper, we have presented the optical design and experimental data for an S2-VHIS with improved field of view. Analysis of the data shows that the system has a resolution of approximately 22 μm axial (FWHM) and 4.5 μm lateral over a field of view of 1.2 mm, which is an order of magnitude larger than previous systems. The axial and lateral resolution of this system can further improved by increasing the angular-spectral selectivity of the gratings. Theoretically, this can be achieved by increasing the thickness. However, in practice, a thicker grating does not necessarily have better selectivity if the light absorption of the holographic material is high. For our S2-VHIS design, we estimate that with small modifications on our recording medium (PQ PMMA) concentration that the axial resolution of ~10 microns can be reached while for SNR > 12dB.
This system simultaneously displays two depth sections separated by ~70 μm, although other configurations with additional sections (five) have been demonstrated by our group . The number of depth section that a S2-VHIS can display depends on the number of holograms. However, the larger the number of holograms the lower the SNR. At present our group is working in the optimization of five depth sections.
This work was supported in part by the National Institutes of Health (CA118167). Part of this work was performed while George Barbastathis was on sabbatical leave from MIT at the School of Engineering and Applied Science at Harvard University.