Parallel-mode scanning optical sectioning using digital Fresnel holography with three-wave interference phase-shifting

Roy Kelner; Joseph Rosen

doi:10.1364/OE.24.002200

1. Introduction

Fresnel incoherent correlation holography (FINCH) has been introduced almost a decade ago as a method for recording Fresnel holograms under spatially incoherent illumination [1]. Utilizing modern electro-optical devices, such as spatial light modulators (SLMs) and digital image sensors, together with phase-shifting techniques [2], FINCH has given a new birth to the classical concept of self-interference-based incoherent holography [3,4] into the era of digital holography. The ability to encode information from a three-dimensional (3-D) scene or object into a single hologram, without special types of illumination (such as a laser), and later focus into various planes through digital reconstruction of the hologram, using a relatively simple - and thus elegant - apparatus, is indeed appealing. The concepts of FINCH have soon been adapted to configurations for natural day-light holography [5,6], synthetic aperture imagers [7,8], adaptive optics [9] and microscopy [10–12].

The reliance of FINCH upon self-interference goes hand in hand with its suitability for recording holograms of fluorescent objects. FINCH also offers enhanced lateral resolution, in the sense that while the cut-off frequency of its modulation transfer function (MTF) is the same as the cut-off frequency of an incoherent imaging system of similar numerical aperture, the shape of the MTF is similar to that of a coherent imaging system, which is uniform throughout the passed frequencies, and does not decay non-zero frequencies like in an incoherent imaging system [13]. This property is entangled with FINCH ability to violate the well-known Lagrange invariant [14–17], a characteristic that is also found in the closely related Fourier incoherent single channel holography (FISCH) method [18]. A recently suggested microscopy method that is based on 180° rotational shearing interferometry, similarly in part to FISCH, also exhibit this type of resolution enhancement [19]; however, unlike FISCH and FINCH, it cannot maintain depth information. Overall, the characteristics of FINCH render it applicable to biological microscopy [20]. A recently developed technique for axial localization has been successfully demonstrated in FINCH, and is capable of real-time tracking of particles [21]. Still, the limited axial resolution of FINCH [16] may diminish its usefulness for examining thick samples.

Lately, a configuration of FINCH that is capable of optical sectioning, for the imaging of thick objects, has been suggested [22]. The solution relies on the integration of a new electro-optical element into FINCH, referred to as a phase pinhole; its role is thoroughly explained in Section 2. By optionally adding point-wise illumination, a complete confocal solution may be formed, following the principles of confocal scanning microscopy [23]. Another confocal configuration of FINCH soon followed [24]. In the latter configuration, a relay system was positioned right after the objective lens of the system, thus forming an intermediate image plane of the target object. Within that plane, a Nipkow spinning disk was inserted to provide optical sectioning before light even reaches the actual holographic interferometer (i.e., before any splitting of waves occurs). In a sense, this may be considered an off the shelf, readily available solution to optical sectioning, but it is a successful testament that a confocal FINCH system can be made competitive with conventional confocal systems in terms of lateral resolution, sectioning performance and applicability. A major drawback of these two solutions for optical sectioning using FINCH is the requirement of a scanning procedure. In the latter case [24], the scanning is mechanical but is rather quick due to the use of the Nipkow spinning disk. In the former, the scanning does not involve any mechanical movements; yet, it was only demonstrated with a rather slow point-by-point scanning [22]. In this paper, a more efficient scanning scheme is presented, where multiple points are scanned in parallel. Furthermore, a generalized phase-shifting procedure between three or more interfering waves for extracting a desired interference pattern between two waves is presented, offering the possibility of overcoming difficulties due to nonindeality of the electro-optical liquid crystal devices in use, such as the limited fill factor of SLMs. Techniques that involve phase-shifting of more than a single wave in processes of multiple wave interference were described in the past (e.g., [25,26]), and a procedure akin to the hereby described procedure have been used by us in a recent publication [27], in the context of coherent holography. However, its generalization and use in the above context of incoherent holography may have special importance. We note that methods of spatially incoherent holography that allow confocal-like optical sectioning without any scanning (for a specific plane of interest) do exist [28–30]. However, unlike [22], they are unfortunately not suitable for fluorescence imaging [30] or situations in which the user does not have control over the illumination.

2. Methods

2.1 Optical sectioning using FINCH

A schematic of an optical sectioning Fresnel incoherent correlation holography (OS-FINCH) setup is presented in Fig. 1. It is assumed that spatially incoherent light is emitted from (or scattered by) the targeted object, meaning that the object can be considered as a collection of many point sources, where the degree of mutual coherence between each pair of points is zero. Under this assumption, each point source only contributes to its own self-interference pattern, and the recorded hologram is a summation over the intensities of all self-interference patterns.

Fig. 1 Schematic of an optical sectioning FINCH setup: L_o, objective lens; L_c, converging lens; P₁ and P₂, polarizers; SLM₁ and SLM₂, spatial light modulators; a_o, point source; a₁ and a₂, point source images.

Download Full Size | PDF

The FINCH setup in Fig. 1 is configured as a common-path single channel interferometer. This interferometer has only a single arm, and interference occurs between waves that travel along a similar direction, in a shared physical space. Consider a single point source a_o within the complete object. Two images of this point source are formed by the interferometer: the image point a₁ is formed by an imaging system that consists of two refractive lenses, the objective lens L_o and the converging lens L_c, and one diffractive lens that is realized using the left-most SLM, SLM₁; the image point a₂ is formed by an imaging system that only consists of the objective lens L_o and the converging lens L_c. The coexistence of these two imaging system is made possible due to the use of a phase-only SLM, which is an electronically controlled birefringence device. Considering linearly polarized light, polarization components oriented along the active axis of the SLM are modulated, whereas orthogonal components, oriented along the non-active axis of the SLM, are not affected by the SLM. In Fig. 1, the polarizer P₁ is set with its transmission axis oriented at a 45° angle to the active axis of the SLM, thereby allowing the simultaneous formation of the two image points a₁ and a₂. Over the image sensor plane, two spherical waves that originate from the point source a_o exist: one diverging from the image point a₁, and one converging towards the image point a₂. According to [13], a complete overlap between the two spherical waves on the image sensor guarantees optimal image resolution. A second polarizer P₂ is positioned in front of the image sensor, with its transmission axis in parallel to the first polarizer P₁, so that the two waves can interfere over the image sensor plane. The intensity of the interference pattern encodes both the point source position in space and the point source intensity.

In order to achieve optical sectioning, a second phase-only SLM, SLM₂, is incorporated into the system, positioned after SLM₁ (Fig. 1). The two SLMs are positioned with their active axes in parallel, meaning that, ideally, SLM₂ does not influence the wave that is converging towards the image point a₂. SLM₂ enables optical sectioning by acting as a pinhole for the image that is formed on its surface. This image is imaged by the two refractive lenses, L_o and L_c, and the diffractive lens displayed on SLM₁. The idea behind the proposed approach of optical sectioning [22] is to preserve all (or most) of the information acquired through the interference pattern between the waves associated with the image points a₁ and a₂ (representing the point source a_o), while eliminating all (or most) of the information from other point sources. This is achieved by using SLM₂ as a phase pinhole. The phase pinhole mask is composed of two components [22]: a small circular region that is set to one of several values of uniform phase modulation, and an axicon that surrounds the phase pinhole. The latter component is used to deflect light away from the digital sensor, as an alternative to actual light blockage that occurs with an ordinary pinhole, while the former is set to three different values of phase modulation, to enable phase-shifting within the effective aperture of the pinhole.

The mechanism of the phase pinhole based optical sectioning is explained with the help of Fig. 2. For simplicity, only three point sources are considered: a_o, b_o and c_o. It is further assumed that the phase pinhole mask is centered along the image point a₁. Let exp(iθ_k)A₁, exp(iθ_k)B₁ and exp(iθ_k)C₁ denote the spherical waves that diverge away from the image points a₁, b₁ and c₁, respectively, over the image sensor plane, where exp(iθ_k) denotes a phase-shifting constant, usually with θ_k = 2π(k – 1) / 3, for k = 1,2,3. Additionally, let A₂, B₂ and C₂ denote the spherical waves that converge towards the image points a₂, b₂ and c₂, respectively, over the image sensor plane. A single recorded FINCH hologram, for the three point sources a_o, b_o and c_o can be described as:

Fig. 2 Principle of optical sectioning using a phase pinhole in FINCH: L_eq, an equivalent lens, representing both the objective lens L_o and the converging lens L_c (Fig. 1); P₁ and P₂, polarizers; SLM₁ and SLM₂, spatial light modulators; a_o, b_o, and c_o, point sources; a₁, b₁, c₁, a₂, b₂, and c₂, point source images.

Download Full Size | PDF

I (x, y; {a_{o}, b_{o}, c_{o}}) = {| e^{i θ_{k}} A_{1} + A_{2} |}^{2} + {| e^{i θ_{k}} B_{1} + B_{2} |}^{2} + {| e^{i θ_{k}} C_{1} + C_{2} |}^{2} .

In a non-sectioning FINCH setup, a phase-shifting procedure is used to extract a final complex-valued hologram that is free from the twin-image and zeroth order terms. The actual phase-shifting is performed using SLM₁ (which, in this case, is the only SLM in the system). The final complex hologram can be described as:

H (x, y; {a_{o}, b_{o}, c_{o}}) = A_{1} A_{2}^{*} + B_{1} B_{2}^{*} + C_{1} C_{2}^{*} .

In an OS-FINCH setup [22], however, the first SLM, SLM₁, is not used for phase-shifting (letting θ_k = 0°), which is now performed at the small circular region of the second SLM, SLM₂. Figure 2 depicts a situation for which the spherical wave A₁ passes entirely within the phase pinhole, the spherical wave B₁ passes entirely outside of the phase pinhole, and the spherical wave C₁ only passes partially within the phase pinhole. Neglecting the impact of the axicon mask, a single recorded FINCH hologram can be described as:

I (x, y; {a_{o}, b_{o}, c_{o}}) = {| e^{i ϕ_{l}} A_{1} + A_{2} |}^{2} + {| B_{1} + B_{2} |}^{2} + {| e^{i ϕ_{l}} S (C_{1}) C_{1} + [1 - S (C_{1})] C_{1} + C_{2} |}^{2} .

where ϕ_l = 2π(l – 1) / 3, for l = 1,2,3, is the phase modulation value within the phase pinhole, and S(C₁) is a support function that indicates the region over the image sensor in which the spherical beam C₁ is phase modulated by the phase pinhole. Following the complete phase-shifting procedure the final FINCH hologram is:

H (x, y; {a_{o}, b_{o}, c_{o}}) = A_{1} A_{2}^{*} + S (C_{1}) C_{1} C_{2}^{*} .

Comparing Eq. (2) and Eq. (4), for the situation described in Fig. 2, reveals that: information regarding the point source a_o is preserved; information regarding the point source b_o is eliminated; information regarding the point source c_o is partly eliminated, since the effective aperture of its hologram is greatly reduced. That is, only a specific point of interest is properly encoded into the hologram in the proposed procedure, while other points are either completely eliminated or are diminished in intensity. The result is an effective optical sectioning, but in order to record an entire 3-D object, a scanning scheme is required. Note that the result shown in Eq. (4) also holds for the case in which the effect of the axicon that surrounds the phase pinhole is considered. The use of the axicon may benefit the recording process by allowing better usage of the dynamic range of the image sensor. Further, it contributes to the elimination of undesired interference processes in which beams that ideally should not exist are involved, as discussed in Section 2.3.

The scanning procedure for a complete object, occupying a 3-D space, is performed through electrical manipulations of the SLM displayed masks, and does not involved any mechanical intervention. For simplicity, it is assumed that the two refractive lenses, L_o and L_c, and the left-most SLM, SLM₁, all lie within a common x-y plane, as shown in Fig. 2. In practice, this is not the case, though it can be achieved using relay systems. The lens L_eq is an equivalent lens that represents the objective lens L_o and the converging lens L_c (Fig. 1), and its focal length f_eq = f_of_c / (f_o + f_c), where f_o and f_c are the focal lengths of the lenses L_o and L_c, respectively. Let d_s denote the distance between the two SLMs, SLM₁ and SLM₂, and f₁ the focal length of the diffractive lens displayed on SLM₁. The scanning procedure can be performed by considering different transverse planes of interest. Let d_o denote the distance between a specific plane of interest and the objective lens. This plane of interest should be imaged to the plane of SLM₂, which is achieved by setting the focal length of the lens displayed on SLM₁ to:

f_{1} = \frac{d_{o} d_{s} f_{eq}}{d_{s} f_{eq} + d_{o} f_{eq} - d_{o} d_{s}} .

During the scanning procedure, the phase pinhole is laterally shifted to various positions within the plane of SLM₂, until the entire area of interest is covered. Reconstruction of a single point source is performed using its corresponding FINCH hologram, and can be done through Fresnel back-propagation to a distance which is dependent upon the distance d_o. The images of a single point source are formed at distances of d₁ and d₂ to the left and right of the image sensor, respectively. It is easy to show these two distance are:

d_{1} = z_{h} - d_{s}, d_{2} = d_{t} - z_{h},

where d_t = f_eqd_o / (d_o – f_eq), and z_h = 2d_sf_c / (d_s + f_c) is the distance between SLM₁ and the image sensor, which assures a complete overlap between a pair of interfering spherical waves on the image sensor, for any point source located at the front focal plane of the objective lens. The reconstruction distance used for the Fresnel back-propagation is:

d_{rec} = \frac{d_{1} d_{2}}{d_{2} + d_{1}} .

Let $H_{x_{o}, y_{o}, d_{o}} (x, y)$ denote a FINCH hologram recorded for a point source located at (x_o,y_o), a distance d_o from the objective lens. Its reconstruction can be calculated as:

h_{x_{o}, y_{o}, d_{o}} (x, y) = H_{x_{o}, y_{o}, d_{o}} (x, y) * Q (1 / d_{rec}),

where Q(s) = exp[iπsλ^–1(x² + y²)] is a quadratic phase function, and * denotes a two-dimensional convolution. In order to form a reconstruction of an entire scanned plane, the reconstructions [Eq. (8)] should be combined. This can be achieved by assigning to each (x,y) coordinate its value from the hologram recorded with the phase pinhole location that corresponds with the desired reconstructed coordinate.

The necessity for a scanning procedure is a main limitation of the above approach, as the scanning is both time demanding and computationally intense. In order to mitigate this burden, a parallelized scanning scheme is hereby proposed. The basic idea is to encode multiple points from a specific plane of interest into a single hologram, rather than just one, by displaying an array of phase pinholes over SLM₂, instead of just a single phase pinhole, as depicted in Fig. 3. Here, square shaped phase pinholes are used, but other shapes (e.g., a circle) may be used as well. The width of a pinhole is represented by w_p, and the distance between pairs of adjacent pinholes (in both x and y directions) is d_p. The width of the scanned area is denoted by w_s, where it is assumed that the scanned area has the form of a square. In a single pinhole scan, the pinhole is laterally shifted throughout the complete scanned area [Fig. 3(a)], requiring at least $⌈ w_{s} / w_{p} ⌉$ different scanning points within each scanned line, and usually double of that in order to ensure sufficient overlap; meaning that scanning of a single plane requires around $4 {⌈ w_{s} / w_{p} ⌉}^{2}$ different complex-valued FINCH holograms, each formed using at least three exposures. In the proposed scanning scheme, however, the amount of required scanning point within each line is reduced to $⌈ (d_{p} + w_{p}) / w_{p} ⌉$ . Taking into account sufficient overlap between scanned locations, the total number of complex FINCH holograms per plane is around $4 {⌈ (d_{p} + w_{p}) / w_{p} ⌉}^{2}$ . The reduction factor is:

R_{f} = \frac{4 {⌈ w_{s} / w_{p} ⌉}^{2}}{4 {⌈ (d_{p} + w_{p}) / w_{p} ⌉}^{2}} \approx {(\frac{w_{s}}{d_{p} + w_{p}})}^{2},

and may be substantial. For example, a typical scan of the area of an entire SLM of 1,000 × 1,000 pixels in size, using a pinhole of 10 pixels in width, may require around 40,000 complex FINCH holograms. A parallel scan using multiple pinholes, with a distance d_p of 90 pixels would only require around 400 complex FINCH holograms, a reduction of two orders of magnitude. However, the sectioning performance is dependent upon the distance d_p, as demonstrated by the experimental results that follow.

Fig. 3 Masks of (a) a single pinhole and (b) a multiple phase pinholes array.

Download Full Size | PDF

The above description of FINCH assumes that the SLMs modulate the polarization components of light that are oriented along their active axis in full. In practice, however, this is not the case. Considering Fig. 2, for example, this implies that the beams that are converging towards the image points to the right of the image sensor, a₂, b₂ and c₂, are mostly linear in polarization along the non-active axis of SLM₂, but not entirely. This has little affect on the end result in a non-sectioning FINCH, and even in a sectioning FINCH, where only a single pinhole is used [22]. However, using a multiple phase pinholes array, as here proposed, pronounces the contribution of these unwanted components, mainly because we get contributions between waves that interfere with themselves. To overcome this unwanted contribution, we suggest a procedure that is hereafter referred to as three-wave interference phase-shifting (TWIPS). The name is derived from the fact that the interference process may now be considered to be between three waves that originate from a single point source, whereas the phase-shifting process enables to extract a desired cross term between two waves.

2.2 Three-wave interference phase-shifting

Consider the interference of two waves, X₁ and X₂, over the digital sensor plane, where one of the waves is phase-shifted:

I_{k} = {| e^{i θ_{k}} X_{1} + X_{2} |}^{2} = {| X_{1} |}^{2} + {| X_{2} |}^{2} + e^{i θ_{k}} X_{1} X_{2}^{*} + e^{- i θ_{k}} X_{1}^{*} X_{2} .

By recording three patterns with different values of θ_k = θ₁,θ₂,θ₃, and calculating [1]:

I = I_{1} (e^{- i θ_{3}} - e^{- i θ_{2}}) + I_{2} (e^{- i θ_{1}} - e^{- i θ_{3}}) + I_{3} (e^{- i θ_{2}} - e^{- i θ_{1}}),

it is easy to show that Eq. (11) results with

I = c X_{1} X_{2}^{*},

where c = 2i[sin(θ₁ – θ₃) + sin(θ₂ – θ₁) + sin(θ₃ – θ₂)], indicating that the twin-image and zero-order terms in Eq. (10) are eliminated by the above procedure.

Consider now a more general case, in which three waves, W₁, W₂ and W₃ are interfering over the digital sensor plane, meaning that

I = {| W_{1} + W_{2} + W_{3} |}^{2} .

We now show that the ability of applying a controllable phase-shift to any two of the three waves can be used to extract any desired cross term between two of the three interfering waves in Eq. (13) (e.g., W₁W₂^*, W₁W₃^* or W₂W₃^*). Without loss of generality, consider the case of controllable phase-shifts applied to W₁ and W₂ so that

I_{k, l} = {| e^{i θ_{k}} W_{1} + e^{i ϕ_{l}} W_{2} + W_{3} |}^{2} .

Let X₁ = W₁ and X₂ = exp(iϕ_l)W₂ + W₃. Based on the procedure described in Eqs. (10)–(12) it is easy to extract the term

I_{l} = X_{1} X_{2}^{*} = W_{1} {(e^{i ϕ_{l}} W_{2} + W_{3})}^{*} .

If the above procedure is performed with two different values of ϕ_l = ϕ₁, ϕ₂ then

W_{1} W_{2}^{*} = \frac{I_{1} - I_{2}}{e^{- i ϕ_{1}} - e^{- i ϕ_{2}}}, W_{1} W_{3}^{*} = \frac{e^{i ϕ_{1}} I_{1} - e^{i ϕ_{2}} I_{2}}{e^{i ϕ_{1}} - e^{i ϕ_{2}}} .

Extraction of the cross term W₂W₃^* can be achieved in a similar fashion by first taking X₁ = W₂ and X₂ = exp(iθ_k)W₁ + W₃. The term X₁X₂^* is then found for two different values of θ_k, each time via phase-shifting over three different values of ϕ_l. In the Appendix, the TWIPS process is described for the case in which a joint phase-shift can be applied simultaneously to two out of three interfering waves, where there is an ability of applying an additional phase-shift to one of these two waves. Then, the TWIPS concept is extended to the more general case of interference between more than three waves. In the following section, the TWIPS concept is described in the context of the example in Fig. 2.

2.3 Three-wave interference in optical sectioning FINCH

In Section 2.1, the phase-shifting procedure in OS-FINCH has been described with the help of Fig. 2, under the assumption that the waves that converge towards the image points to the right of the image sensor are not modulated by SLM₂, upon which the phase pinhole mask is displayed. Even though this is mostly satisfied, in practice, SLM₁ does not modulate the waves that have their polarization components along its active axis in full. This may result, in part, from the limited fill factor of currently available devices. Thus, the waves that converge towards the image point to the right of the image sensor also possess components along the active axis of SLM₂. Even though these components may be relatively small, a way to mitigate their influence should be considered, especially when multiple phase pinholes are used.

Consider the point source c_o in Fig. 4. The interference pattern due to this point source over the image sensor includes several waves. The wave that diverges from the image point c₁, is modulated by both SLMs and can be described as exp(iφ_k){exp(iϕ_l)S(C₁)C₁ + [1 – S(C₁)]C₁}, where φ_k and ϕ_l are phase-shifting parameters controlled via SLM₁ and SLM₂, respectively. The latter is only set within the phase pinhole regions on SLM₂ [Fig. 3], while the former is set on SLM₁ entirely. The support function S(C₁) is defined in Section 2.1 following Eq. (3). The wave that converges towards the image point c₂, has dominant polarization components along the non-active axis of SLM₂, which can be denoted as C₂, and undesired components along the active axis of SLM₂, described as exp(iϕ_l)S(C₃)C₃ + [1 – S(C₃)]C₃, where part of the wave is phase-shifted when passed through the phase pinhole, as indicated by the support function S(C₃). Other components that may exist are assumed to be negligible and are neglected from the analysis. Thus, the recorded interference pattern for the point source c_o can be written as:

Fig. 4 Wave interference in optical sectioning FINCH: L_eq, an equivalent lens, representing both the objective lens L_o and the converging lens L_c (Fig. 1); P₁ and P₂, polarizers; SLM₁ and SLM₂, spatial light modulators; c_o, a point source; c₁ and c₂, point source images; C₁, the wave that diverges from the image point c₁ over the image sensor plane, composed of parts that pass through the phase pinhole [S(C₁)C₁] and parts that do not {[1 – S(C₁)C₁}; C₂, the wave that converges towards the image point c₂ over the image sensor plane and is not modulated by SLM₂; C₃, the wave that converges towards the image point c₂ over the image sensor plane and is modulated by SLM₂, composed of parts that pass through the phase pinhole [S(C₃)C₃] and parts that do not {[1 – S(C₃)C₃}. For simplicity, only a single phase pinhole is shown.

Download Full Size | PDF

I_{k, l} (x, y; {c_{o}}) = {| e^{i φ_{k}} {e^{i ϕ_{l}} S (C_{1}) C_{1} + [1 - S (C_{1})] C_{1}} + C_{2} + e^{i ϕ_{l}} S (C_{3}) C_{3} + [1 - S (C_{3})] C_{3} |}^{2}

The form of Eq. (17) may be considered as the interference of five waves. Taking into account that the axicon displayed over SLM₂ deflects two of them from reaching the image sensor, Eq. (17) can be reduced to:

I_{k, l} (x, y; {c_{o}}) = {| e^{i (ϕ_{l} + φ_{k})} S (C_{1}) C_{1} + e^{i ϕ_{l}} S (C_{3}) C_{3} + C_{2} |}^{2} .

Note that the form of the above equation is similar in form to Eq. (19) in the Appendix. Therefore, the TWIPS procedure described in Eq. (20), Eq. (21) and the left-hand-side of Eq. (22) can be used to extract the cross term S(C₁)C₁C₂^* using a total of six exposures. This is the exact same term that represents the point source c_o in Eq. (4), indicating that the suggested method can effectively deal with the undesired waves that are included in the analysis.

3. Experiments

An experimental setup based on the schematic of Fig. 5 was implemented. It is based directly upon the configuration of Fig. 1, but uses reflective SLMs (Holoeye PLUTO, 1920 × 1080 pixels, 8 μm pixel pitch, phase-only modulation) instead of transmissive ones. A beam splitter was inserted next to each of the SLMs, so that they can be used like transmissive type SLMs. For simplicity, the two lenses L_o and L_c were replaced by a single equivalent lens L_eq of focal length f_eq = 20 cm, representing an objective lens L_o of focal length f_o = 23.08 cm and a converging lens L_c of focal length f_c = 150 cm. The distance between the lens L_eq and SLM₁ was set to 10 cm. SLM₂ and the image sensor (Hamamatsu ORCA-Flash4.0 V2 Digital CMOS, 2048 × 2048 pixels, 6.5μm pixel pitch, monochrome) were positioned at distances of 66.32 cm and 90 cm away from SLM₁, respectively. Two resolution charts were combined using a beam combiner into a 1 cm thick object. A negative 1951 United States Air Force (USAF) resolution chart was placed at a distance f_o in front of the lens L_eq, while a negative National Bureau of Standards (NBS) 1963A was placed at a distance of f_o + 1 cm = 24.08 cm in front of the same lens. Each resolution chart was back-illuminated using an LED (Thorlabs LED635L, 170 mW, central wave length of λ = 635 nm, full width at half maximum Δλ = 15 nm). The exposure time of the image sensor was set to 80 ms per grabbed frame.

Fig. 5 Optical sectioning FINCH experimental setup: L_eq, an equivalent lens, representing both the objective lens L_o and the converging lens L_c (Fig. 1); P₁ and P₂, polarizers; SLM₁ and SLM₂, spatial light modulators; RC₁ and RC₂, resolution charts; BS₁, BS₂ and BS₃, beam splitters (BS₁ is used as a beam combiner); LEDs, light emitting diodes; L_is, illumination lenses.

Download Full Size | PDF

At first, regular (non-sectioning) FINCH holograms were recorded. The mask of the second SLM, SLM₂, was set to a uniform phase-modulation value of zero, electronically transforming the setup into a regular FINCH configuration. Two FINCH holograms were recorded, each calculated according to Eq. (11) from three exposures with difference phase-shifting values of θ₁ = 0°, θ₂ = 120° and θ₃ = 240°. The first and second holograms were recorded with the USAF and NBS targets imaged upon SLM₂, respectively. Reconstruction results from the first hologram, showing the USAF and NBS charts in planes of best focus are shown in Figs. 6(a) and 6(b), respectively. Note that details of the two charts are not easily resolved in regions within the two charts overlap. In the case of the second hologram, the USAF and NBS charts appear in best focus simultaneously, as shown in the reconstruction in Fig. 6(c). These results are brought here to allow comparison with the results from an optical sectioning capable FINCH setup that follow.

Fig. 6 Reconstructions of non-sectioning FINCH holograms at plane of best focus for 1951 USAF chart [(a) and (c)] and NBS 1963A chart [(b) and (c)]. (a) and (b) were reconstructed from the first hologram, recorded with the 1951 USAF chart imaged to the plane of SLM₂. (c) was reconstructed from the second hologram, recorded with the NBS 1963A chart imaged to the plane of SLM₂.

Download Full Size | PDF

In the optical sectioning experiments, each of the two planes of interest (the two resolution charts) were first scanned using a square shaped phase pinhole of 88 μm in width (i.e., 11 pixels over 11 pixels in size). The scan was performed on a grid of size 101 × 101 points, with the phase pinhole shifted in jumps of 40 μm, enabling sufficient overlap between imaged positions. For each pinhole position, a complex FINCH hologram was calculated from six different exposures, based on the procedure described in Section 2.3, where SLM₁ was set with φ₁ = 0°, letting the phase pinhole, which is displayed on SLM₂, take the values ϕ₁ = 0°, ϕ₂ = 120° and ϕ₃ = 240°, or with φ₂ = 180°, again letting the phase pinhole take the three values of ϕ₁ = 0°, ϕ₂ = 120° and ϕ₃ = 240°. Then, scanning using multiple phase pinholes, in parallel, was tested. The spacing between adjacent phase pinholes (in both x and y direction) was set to values of 88 × g μm, with g = 1 .. 15, implying spacing of 1 to 15 pinholes in width, between pinholes. It should be emphasized that: (1) a trial to perform a regular phase-shifting of three exposures, according to [22], does not yield acceptable results when multiple phase pinholes are displayed; (2) a mask of an axicon that occupy the phase pinhole(s) surrounding(s) was indeed used in the presented experiments and yielded better results than the case of working without the axicon; (3) the exposure time of 80 ms per grabbed frame indicates that a single complex FINCH hologram requires roughly 0.5 s for its recording. In practice, the total time required for the recording was about 2 s, a figure which can be reduced with tighter synchronization.

Reconstruction results, based on the procedure described in Section 2, for the holograms recorded with the 1951 USAF resolution chart as the plane of interest, are presented in Fig. 7. Visual comparison of the reconstruction from the hologram set acquired using only a single phase pinhole [Fig. 7(a)] with the results of multiple phase pinholes, starting with the spacing of a single pinhole in Fig. 7(b) through spacing of 15 pinholes in Fig. 7(l), reveals gradual improvement in the sectioning performance that increases with the spacing between the pinholes, until a spacing of 7 pinholes is reached [Fig. 7(h)]. From that point, any increase in the spacing between pinholes does not reveal visual improvement in the sectioning results [Figs. 7(h)–7(l)], which resemble the performance in the case of a single pinhole [Fig. 7(a)]. That is, there is a trade-off between the complexity of the scan and the sectioning performance. We note that by visual comparison there is a clear improvement even between the highly parallelized (with g = 1) optical sectioning FINCH in Fig. 7(b) and the non-sectioning FINCH in Fig. 6(a). To quantify the results, the mean squared error (MSE) was calculated between the amplitude of the reconstruction in Fig. 7(a) and the amplitude of the reconstructions in Figs. 7(b)–7(l). The cases of g = 11 .. 14, which are not presented in Fig. 7, for brevity, were also included in the MSE calculations. Further, the non-sectioning FINCH reconstruction in Fig. 6(a) is considered to represent the case of g = 0. The results are presented in Fig. 8, and are fairly consistent with the visual inspection. Taking into consideration that Fig. 7(a) was generated using 10,102 complex FINCH holograms, Fig. 7(i) using more than an order of a magnitude less (400 complex holograms), and Fig. 7(b) using only 25 complex holograms, it is clear that in practice it may be necessary to balance between the required sectioning and the time that is available for the process (here, the total time taken for scanning a single plane varied between few seconds to several hours, depending on the value of g). Still, the scanning time may be substantially decreased, without impairing the sectioning performance. Reconstruction results for the NBS 1963A resolution chart are presented in Fig. 9, with the MSE graph presented in Fig. 10. The previous conclusions are further supported by the results.

Fig. 7 Reconstructions of optical sectioning FINCH holograms at plane of best focus for the 1951 USAF chart. In (a) a single phase pinhole was used, while in (b)–(l) multiple phase pinholes were used in parallel, where the spacing between adjacent pinholes (in x and y directions) was set to 88 × g μm, starting with g = 1 in (b) and ending with g = 15 in (l). Results for g = 11 .. 14 are not presented, due to lack of visible difference.

Download Full Size | PDF

Fig. 8 Amplitude mean squared error (MSE) between the reconstruction in Fig. 7(a), where scanning was performed using a single pinhole, and the reconstructions in Figs. 7(b) through 7(l), where parallel mode scanning was performed using multiple phase pinholes, with g values of 1 through 15, respectively, indicating the gap between adjacent phase pinhole, in integer multiplies of the width of a single pinhole. The result for g value of 0 is against the reconstruction of Fig. 6(a), from a non-sectioning FINCH hologram.

Download Full Size | PDF

Fig. 9 Reconstructions of optical sectioning FINCH holograms at plane of best focus for the NBS 1963A chart. In (a) a single phase pinhole was used, while in (b)–(l) multiple phase pinholes were used in parallel, where the spacing between adjacent pinholes (in x and y directions) was set to 88 × g μm, starting with g = 1 in (b) and ending with g = 15 in (l). Results for g = 11 .. 14 are not presented, due to lack of visible difference.

Download Full Size | PDF

Fig. 10 Amplitude mean squared error (MSE) between the reconstruction in Fig. 9(a), where scanning was performed using a single pinhole, and the reconstructions in Figs. 9(b) through 9(l), where parallel mode scanning was performed using multiple phase pinholes, with g values of 1 through 15, respectively, indicating the gap between adjacent phase pinhole, in integer multiplies of the width of a single pinhole. The result for g value of 0 is against the reconstruction of Fig. 6(c), from a non-sectioning FINCH hologram.

Download Full Size | PDF

Finally, to demonstrate the necessity of the TWIPS procedure, Fig. 11 presents the reconstructions of two different complex FINCH hologram, acquired with g values of 15 and 1, starting with a relatively low number of in-parallel phase pinholes, and ending with a relatively large number. The reconstructions of the complex FINCH holograms, using a regular phase-shifting procedure with three different exposures, are presented in Figs. 11(a) and 11(b). Besides showing small fractions of the 1951 USAF resolution chart that fall within the phase pinholes, noise-like artifacts are also visible in the reconstructions. The higher the number of in-parallel phase pinholes (indicated by lower values of g), the more prominent are the artifacts. Using TWIPS, however, where each complex FINCH hologram requires a total of six different exposures, these artifacts are eliminated, as shown in Figs. 11(c) and 11(d). Thus, the contribution of TWIPS is apparent, and its necessity for the successful utilization of the suggested parallel scanning scheme in OS-FINCH is demonstrated, especially when the degree of parallelism is high.

Fig. 11 Regular phase-shifting vs. TWIPS in OS-FINCH. In (a) and (b) hologram reconstructions from complex FINCH holograms are shown, each calculated using a regular phase-shifting procedure, based on three different exposures. Each of the holograms was recorded for a single position of multiple phase pinholes, with the distance between adjacent phase pinholes set to 88 × g μm, with g = 15 (a) and g = 1 (b). (c) and (d) are the TWIPS equivalents of (a) and (b), respectively, where each complex FINCH hologram was calculated based on six different exposures. Using regular phase-shifting, it is evident that the larger the number of phase pinholes, comparing (a) and (b), the more prominent are the noise-like artifacts in the reconstructions. In contrast, these artifacts are not noticed in the TWIPS reconstructions of (c) and (d).

Download Full Size | PDF

4. Summary and conclusions

A new scanning procedure in an optical sectioning FINCH configuration has been presented. The procedure offers parallel-mode scanning capable of imaging a large number of points in an examined object plane simultaneously. This is achieved through the use of multiple phase pinholes. The speed gain is dependent upon the specific characteristics of the object. In the presented experimental results, a speedup of roughly one order of magnitude has been achieved, without any deterioration of the sectioning capabilities. Higher speedups are possible, but require a balance between the desired speed and the quality of the sectioning.

The use of multiple phase pinholes based scan is more sensitive to some of the nonidealities of currently available SLM devices, in comparison to the situation of using just a single phase pinhole. The suggested analysis of the case suggests that for a single point source, more than two waves are involved in the interference process. This is dealt with through a phase-shifting procedure which is referred to as TWIPS. The procedure effectively eliminates any of the unwanted interference terms, offering successful implementation of the suggested parallel-mode scan. TWIPS is discussed not only in the context of FINCH, but also in a broader and more general sense, so that it may readily be adapted to other systems as well, if necessary.

Overall, in OS-FINCH the sectioning is done only in the image plane, where the phase pinholes operate. Therefore, on the one hand, the sectioning performance of OS-FINCH are expected to be inferior to that of a conventional confocal microscope. This conclusion is valid at least until a confocal version of the proposed OS-FINCH system is demonstrated, in which a matching pinhole array will be used to illuminate the speciemen. On the other hand, it should be noted that the present OS-FINCH is suitable to imaging situations in which there is no access or possibility to illuminate speciemens with a beam array, like in confocal microscopy. Another advantage of OS-FINCH over the common confocal microscopes equipped with a spinning disk [24], is that in the OS-FINCH there are not any moving parts. The entire 3-D scanning is electronic rather than mechanical, and therefore the scanning may be potentially faster and more robust.

Appendix

The case in which three waves, W₁, W₂ and W₃, are interfering over the digital sensor plane is considered in this appendix. This time, it is assumed that a simultaneous phase-shift can be applied to two of these waves, say W₁ and W₂, and that an additional phase-shift can be separately applied to W₁. This is in contrast to the previously discussed case, in which two different phase-shifts could be applied separately to W₁ and W₂. Thus, the intensity of the interference between the waves is:

I_{k, l} = {| e^{i (ϕ_{l} + φ_{k})} W_{1} + e^{i ϕ_{l}} W_{2} + W_{3} |}^{2},

where the phase-shifting parameter ϕ_l accounts for both W₁ and W₂ and the parameter φ_k to W₁ alone. Define φ_k = θ_k – ϕ_l, Eq. (19) becomes similar to Eq. (14), and thus any desired cross term between the three interfering waves can be extracted, following Eqs. (14)–(16). This definition of φ_k can be considered as a way to compensate for any value of ϕ_l, effectively allowing separate phase-shifting control over W₁ and W₂. An alternative procedure that does not use this technique of compensation is now presented.

Let θ_l = ϕ_l, X₁ = exp(iφ_k)W₁ + W₂ and X₂ = W₃. Equation (19) can be rewritten as:

I_{k, l} = {| e^{i θ_{l}} X_{1} + X_{2} |}^{2} = {| X_{1} |}^{2} + {| X_{2} |}^{2} + e^{i θ_{l}} X_{1} X_{2}^{*} + e^{- i θ_{l}} X_{1}^{*} X_{2} .

Following a procedure similar to the one described in Eqs. (10)–(12), phase-shifting over three different values of θ_l, it is easy to extract from Eq. (20) the cross term

I_{k} = X_{1} X_{2}^{*} = (e^{i φ_{k}} W_{1} + W_{2}) W_{3}^{*} .

Performing the above for two different values of φ_k = φ₁, φ₂, the following cross terms of Eq. (19) can be extracted:

W_{1} W_{3}^{*} = (I_{1} - I_{2}) / (e^{i φ_{1}} - e^{i φ_{2}}), W_{2} W_{3}^{*} = (e^{- i φ_{1}} I_{1} - e^{- i φ_{2}} I_{2}) / (e^{- i φ_{1}} - e^{- i φ_{2}}) .

Alternatively, let θ_k = φ_k, X₁ = exp(iϕ_l)W₁ and X₂ = exp(iϕ_l)W₂ + W₃. Equation (19) can be rewritten as:

I_{k, l} = {| e^{i θ_{k}} X_{1} + X_{2} |}^{2} = {| X_{1} |}^{2} + {| X_{2} |}^{2} + e^{i θ_{k}} X_{1} X_{2}^{*} + e^{- i θ_{k}} X_{1}^{*} X_{2} .

Phase-shifting over three different values of θ_k, it is easy to extract from Eq. (23) the cross term

I_{l} = X_{1} X_{2}^{*} = e^{i ϕ_{l}} W_{1} {(e^{i ϕ_{l}} W_{2} + W_{3})}^{*} = W_{1} {(W_{2} + e^{- i ϕ_{l}} W_{3})}^{*} .

Implementing the above for two different values of ϕ_l = ϕ₁, ϕ₂, the following cross terms of Eq. (19) can be extracted:

W_{1} W_{2}^{*} = (e^{- i ϕ_{1}} I_{1} - e^{- i ϕ_{2}} I_{2}) / (e^{- i ϕ_{1}} - e^{- i ϕ_{2}}), W_{1} W_{3}^{*} = (I_{1} - I_{2}) / (e^{i ϕ_{1}} - e^{i ϕ_{2}}) .

Finally, consider a more general case, in which multiple waves, W₁, W₂ .. W_p, are interfering over the digital sensor plane, so that

I = {| W_{1} + W_{2} + .. + W_{p} |}^{2} .

We now show that the ability of applying a controllable phase-shift to two of the waves, say W_m and W_n from W₁, W₂ .. W_p, can be used to extract a desired cross term,

W_{m} W_{n}^{*}

, from Eq. (26), so that

I_{k, l} = {| W_{1} + .. + W_{m - 1} + e^{i θ_{k}} W_{m} + W_{m + 1} + .. + W_{n - 1} + e^{i ϕ_{l}} W_{n} + W_{n + 1} + .. + W_{p} |}^{2} .

Let X₁ = W_m and X₂ = W₁ + .. + W_m_–1 + W_m₊₁ + .. + exp(iϕ_l)W_n + .. + W_p. Equation (27) can be rewritten in the form of Eq. (23). Phase-shifting over three different values of θ_k, based on the procedure described in Eqs. (10)–(12) it is easy to extract the term

I_{l} = X_{1} X_{2}^{*} = W_{m} {(W_{1} + .. + W_{m - 1} + W_{m + 1} + .. + e^{i ϕ_{l}} W_{n} + .. + W_{p})}^{*} .

If the above procedure is performed with two different values of

ϕ_{l} = ϕ_{1}, ϕ_{2}

then

I = (I_{1} - I_{2}) / (e^{- i ϕ_{1}} - e^{- i ϕ_{2}}) = W_{m} W_{n}^{*} .

Acknowledgment

This work was supported by The Israel Ministry of Science, Technology and Space and by The Israel Science Foundation (ISF) (Grant No. 439/12). JR was supported by a research fellowship granted by the University of Electro-Communications (UEC), Japan.

References and links

1. J. Rosen and G. Brooker, “Digital spatially incoherent Fresnel holography,” Opt. Lett. 32(8), 912–914 (2007). [CrossRef] [PubMed]

2. I. Yamaguchi and T. Zhang, “Phase-shifting digital holography,” Opt. Lett. 22(16), 1268–1270 (1997). [CrossRef] [PubMed]

3. A. W. Lohmann, “Wavefront reconstruction for incoherent objects,” J. Opt. Soc. Am. 55(11), 1555–1556 (1965). [CrossRef]

4. G. Cochran, “New method of making Fresnel transforms with incoherent light,” J. Opt. Soc. Am. 56(11), 1513–1517 (1966). [CrossRef]

5. M. K. Kim, “Full color natural light holographic camera,” Opt. Express 21(8), 9636–9642 (2013). [CrossRef] [PubMed]

6. O. Cossairt, N. Matsuda, and M. Gupta, “Digital refocusing with incoherent holography,” in Proceedings of IEEE International Conference on Computational Photography (ICCP, 2014).

7. B. Katz and J. Rosen, “Could SAFE concept be applied for designing a new synthetic aperture telescope?” Opt. Express 19(6), 4924–4936 (2011). [CrossRef] [PubMed]

8. Y. Kashter, Y. Rivenson, A. Stern, and J. Rosen, “Sparse synthetic aperture with Fresnel elements (S-SAFE) using digital incoherent holograms,” Opt. Express 23(16), 20941–20960 (2015). [CrossRef] [PubMed]

9. M. K. Kim, “Adaptive optics by incoherent digital holography,” Opt. Lett. 37(13), 2694–2696 (2012). [CrossRef] [PubMed]

10. J. Rosen and G. Brooker, “Non-scanning motionless fluorescence three-dimensional holographic microscopy,” Nat. Photonics 2(3), 190–195 (2008). [CrossRef]

11. P. Bouchal and Z. Bouchal, “Wide-field common-path incoherent correlation microscopy with a perfect overlapping of interfering beams,” J. Eur. Opt. Soc: Rapid Pub. 8, 13011 (2013).

12. T. Yanagawa, R. Abe, and Y. Hayasaki, “Three-dimensional mapping of fluorescent nanoparticles using incoherent digital holography,” Opt. Lett. 40(14), 3312–3315 (2015). [CrossRef] [PubMed]

13. J. Rosen, N. Siegel, and G. Brooker, “Theoretical and experimental demonstration of resolution beyond the Rayleigh limit by FINCH fluorescence microscopic imaging,” Opt. Express 19(27), 26249–26268 (2011). [CrossRef] [PubMed]

14. P. Bouchal, J. Kapitán, R. Chmelík, and Z. Bouchal, “Point spread function and two-point resolution in Fresnel incoherent correlation holography,” Opt. Express 19(16), 15603–15620 (2011). [CrossRef] [PubMed]

15. X. Lai, S. Zeng, X. Lv, J. Yuan, and L. Fu, “Violation of the Lagrange invariant in an optical imaging system,” Opt. Lett. 38(11), 1896–1898 (2013). [CrossRef] [PubMed]

16. J. Rosen and R. Kelner, “Modified Lagrange invariants and their role in determining transverse and axial imaging resolutions of self-interference incoherent holographic systems,” Opt. Express 22(23), 29048–29066 (2014). [CrossRef] [PubMed]

17. X. Lai, S. Xiao, Y. Guo, X. Lv, and S. Zeng, “Experimentally exploiting the violation of the Lagrange invariant for resolution improvement,” Opt. Express 23(24), 31408–31418 (2015). [CrossRef] [PubMed]

18. R. Kelner, J. Rosen, and G. Brooker, “Enhanced resolution in Fourier incoherent single channel holography (FISCH) with reduced optical path difference,” Opt. Express 21(17), 20131–20144 (2013). [CrossRef] [PubMed]

19. D. Weigel, H. Babovsky, A. Kiessling, and R. Kowarschik, “Widefield microscopy with infinite depth of field and enhanced lateral resolution based on an image inverting interferometer,” Opt. Commun. 342, 102–108 (2015). [CrossRef]

20. G. Brooker, N. Siegel, J. Rosen, N. Hashimoto, M. Kurihara, and A. Tanabe, “In-line FINCH super resolution digital holographic fluorescence microscopy using a high efficiency transmission liquid crystal GRIN lens,” Opt. Lett. 38(24), 5264–5267 (2013). [CrossRef] [PubMed]

21. P. Bouchal and Z. Bouchal, “Non-iterative holographic axial localization using complex amplitude of diffraction-free vortices,” Opt. Express 22(24), 30200–30216 (2014). [CrossRef] [PubMed]

22. R. Kelner, B. Katz, and J. Rosen, “Optical sectioning using a digital Fresnel incoherent-holography-based confocal imaging system,” Optica 1(2), 70–74 (2014). [CrossRef] [PubMed]

23. M. Minsky, “Memoir on inventing the confocal scanning microscope,” Scanning 10(4), 128–138 (1988). [CrossRef]

24. N. Siegel and G. Brooker, “Improved axial resolution of FINCH fluorescence microscopy when combined with spinning disk confocal microscopy,” Opt. Express 22(19), 22298–22307 (2014). [CrossRef] [PubMed]

25. P. K. Rastogi, “Phase shifting applied to four-wave holographic interferometers,” Appl. Opt. 31(11), 1680–1681 (1992). [CrossRef] [PubMed]

26. M. A. Volynsky and I. P. Gurov, “Investigation of three-beam interference fringes with controllable phase shift of two reference waves,” Proc. SPIE 7790, 779015 (2010). [CrossRef]

27. E. Shtraikh, M. Zieder, R. Kelner, and J. Rosen, “In-line digital holography using adaptive phase pinhole on the Fourier plane,” Asian J. Phys.24(12) 2015, in press.

28. P.-C. Sun and E. N. Leith, “Broad-source image plane holography as a confocal imaging process,” Appl. Opt. 33(4), 597–602 (1994). [CrossRef] [PubMed]

29. P. C. Sun and E. Arons, “Nonscanning confocal ranging system,” Appl. Opt. 34(7), 1254–1261 (1995). [CrossRef] [PubMed]

30. R. Chmelík and Z. Harna, “Parallel-mode confocal microscope,” Opt. Eng. 38(10), 1635–1639 (1999). [CrossRef]

Parallel-mode scanning optical sectioning using digital Fresnel holography with three-wave interference phase-shifting

Abstract

1. Introduction

2. Methods

2.1 Optical sectioning using FINCH

2.2 Three-wave interference phase-shifting

2.3 Three-wave interference in optical sectioning FINCH

3. Experiments

4. Summary and conclusions

Appendix

Acknowledgment

References and links

Cited By

Figures (11)

Equations (29)

Optics Express