Resonant Scanning in Laser Confocal Microscopy
Laser scanning confocal microscopy has proven to be a useful tool for examining fixed and stained cells, tissues, and even whole organisms at high contrast by the elimination of light originating in regions removed from the focal plane. The growing application of fluorescent proteins in live-cell imaging, however, now requires microscope imaging speeds on the millisecond timescale in order to unravel the intricate dynamics that occur in many biological processes. Unfortunately, traditional laser scanning confocal microscopes are limited in acquisition speed by the galvanometer mirrors, which are driven with a linear saw-tooth control signal at the rate of several microseconds per pixel. This translates to a scan rate ranging from 500 milliseconds to 2 seconds, depending upon the image dimensions. In order to acquire images on faster timescales, laser scanning confocal microscopes must be re-engineered to incorporate advanced scanning scenarios that enable the beam to be raster-scanned across the specimen at higher speeds. To overcome the inherently slow speed of confocal microscopes, several manufacturers have introduced instruments equipped with resonant scanning mirrors that are capable of gathering images at 30 frames per second or higher.
Illustrated in Figure 1 is a Nikon A1R laser confocal microscope scanning unit equipped with a number of advanced features, including a high-resolution galvanometer-based scanner (4,096 x 4,096 pixels; non-resonant) and a high-speed resonant scanner, a continuously variable hexagonal pinhole, spectral imaging capacity, several optical output ports, and a pair of input ports for multiple lasers of different wavelengths. An additional free space beam introduction port is included in the A1R-MP (multiphoton model) for the ultrafast laser necessary for multiphoton imaging. The resonant scanning system is capable of high speed image capture at rates ranging from 30 frames per second (512 x 512 pixels) to 420 frames per second (152 x 32 pixels) whereas the non-resonant scanner features a maximum scan rate of 4 frames per second at 512 x 512 pixels. Confocal zoom with the resonant scanner (from 1.5x to 8x) is limited by the requirement of a separate Ronchi grating pattern for each step (discussed below), whereas the non-resonant scanner features a continuously variable zoom range of 1x to 1000x. Perhaps the most advanced feature of the A1R scanning unit is the ability to perform hybrid scanning using both scanners in tandem. In this mode, the non-resonant scanner can be used for photoactivation or photobleaching of selected regions of the specimen followed by subsequent high-speed imaging with the resonant scanner to observe recovery or diffusion of the photoactivated species in real time. The true power of this instrument lies in the multiple features that enable both high resolution confocal imaging for fixed specimens and real-time scanning with optional photoactivation for live-cell imaging using fluorescent proteins, optical highlighters, and caged fluorophores. In microscope core facilities with a limited budget, such versatility should enable the accommodation of a wider user base in a single instrument.
Although relatively sophisticated resonant galvanometers have been commercially available for more than three decades, their application in laser scanning confocal microscopy has yet to experience widespread utility. The first point-scanning instruments were fabricated in the laboratories of investigators wishing to image fast calcium waves in living cells during the early 1990s. After several research reports describing successful experiments emerged, other researchers began to develop more sophisticated instruments, including multiphoton confocals containing resonant scanners. Construction of resonant scanning microscopes is complicated by the cosinusoidal motion of the high-speed resonant scanner, which creates problems with pixel clocking. In the past decade, however, several workable solutions (both hardware and software based) have emerged that enable the instruments to perform normal confocal functions, such as pan and zoom, at video rates. These instruments should enjoy increased popularity in the upcoming years and will undoubtedly be able to compete with spinning disk and line scanners in both speed and imaging efficiency.
In a traditional laser scanning confocal microscope, the excitation light beam is expanded and directed to a pair of oscillating galvanometer scanning mirrors that raster-scan the focused beam across the specimen. Fluorescence emission is de-scanned through the same mirror set and passed through a conjugate (confocal) pinhole aperture to remove out-of-focus light before entering the photomultiplier detector. During the scanning and de-scanning process, one of the galvanometer mirrors scans the specimen along the fast, horizontal axis, while the other mirror scans and tracks the slower vertical axis. Scanning continues until an entire two-dimensional (x-y) image is gathered, a process that can be repeated to generate a series of images over time. Alternatively, the focal point can be stepped along the microscope axial plane to acquire a three-dimensional (x, y, and z) image stack of optical sections. The scanning speed is limited by the mechanical specifications of the fast-axis mirror, which typically scans at a rate of approximately 4 to 5 microseconds per pixel. Thus, for a 512 x 512 pixel image collected in a single second, the scanning spot dwells on each pixel for about 4 microseconds. Driving a standard galvanometer scanner at video rates (30 frames per second) to gather an image of similar size is difficult, if not impossible, due to the fact that the mirror would have to be rapidly accelerated, held at a constant velocity while scanning across the field, then rapidly decelerated and the direction of travel reversed, repeating this cycle for each scanned line. Attempting such an action at 30 frames per second would create a condition that could lead to overheating and premature failure. As a result, the slow scanning rates necessitated by non-resonant linear galvanometers limit the ability of confocal microscopes to observe events that occur on very fast timescales.
Approaches to Fast Specimen Scanning
With galvanometer-based laser scanning confocal microscopes, the fastest scanning scenario involves using line scans to acquire a row of pixels along a single axis of the specimen with the fast-axis mirror while relegating the slow-axis mirror to selecting the appropriate location for scanning (Figure 2). In modern confocal instruments, the scan line is not constrained to the x-axis, but can be oriented in any lateral direction to include all specimen features of interest. The results are often plotted as a series of scans that are offset to enable visualization of fluorescence intensity changes with time. This approach can produce high-speed scans, but is limited in its ability to detect events that might occur in other regions of the specimen. Another pitfall of fast line-scanning is that the dwell time on each pixel is generally insufficient to excite all of the detectable fluorescence, especially with specimens that have poor target abundance. In addition, the photomultipliers used in confocal microscopy have a quantum efficiency ranging from approximately 15 to 40 percent, so as scanning speeds increase, the number of detected photons per pixel may drop in some specimens to a level that is obscured by the noise floor of the photomultiplier.
Presented in Figure 2 are examples of line-scanning confocal microscopy in living cells. The sequence shown in Figure 2(a) captures an xt-scan series of calcium sparks induced by the synthetic indicator reagent, Fura Red, in cardiac muscle cells. Note the regions of high fluorescence intensity that occur in several locations during successive time points in the sequence. In Figure 2(b), images of calcium waves are illustrated traversing the cytoplasm in an isolated HeLa cell using a fluorescent protein cameleon biosensor probe (containing mCerulean and mVenus). Obtaining images of high spatial resolution (such as 512 x 512 pixels) requires fast x-y scanning and must be conducted with a microscope that can sweep the entire field in milliseconds, exemplified by the instruments described in the following paragraphs. Using rapid line scanning with a linear-galvanometer equipped confocal microscope, the calcium wave can be recorded in one dimension as a function of time as illustrated in the graph below the image set. In Figure 2(c), scans were conducted along the white dotted line in Figure 2(b) to capture a single spatial dimension of the wave, which passes through the cell in less than a second.
Another strategy for fast image acquisition involves increasing the pixel dwell time without lowering the frame rate by shaping the illumination beam into a line that simultaneously excites the fluorophores across one axis, while scanning the specimen along the other axis to achieve video frame rates at moderate mirror speeds (schematically illustrated in Figure 3(c)). This scanning concept has been implemented in a commercial confocal instrument that eliminates the fast-axis oscillating galvanometer mirror and captures fluorescence emission using a linear CCD array (line-CCD). Thus, a line-scanning confocal microscope can achieve video frame rates by exciting with line-shaped illumination and gathering fluorescence emission through a confocal slit aperture. The practical limitation of line-scanning confocal instruments is the asymmetrical resolution that is obtained due to the fact that resolution in one axis is provided by confocal scanning whereas resolution in the other axis is limited by conventional widefield optics. In addition, it is technically difficult to generate a narrow diffraction-limited line of focused light coupled with the appropriately sized detection slit at resolutions that are not substantially lower than the theoretical optimum.
Commercial slit confocal microscope solutions rely on using slit apertures of different widths (to vary the degree of confocality) that are fabricated by depositing a thin film of opaque material on optical glass. The scanning illumination is a wedge of light formed by a cylindrical lens element that converges on the focal plane at an angle determined by the numerical aperture of the objective. Slit scanning instruments are capable of high speed image acquisition at resolutions that are hampered only by a slight degradation of the image point spread function in the axial direction, as mentioned above. On the downside, slit scanners currently rely on linear one-pixel CCD detectors that have similar quantum efficiency and noise characteristics to scientific-grade CCDs. These detectors are far less sensitive than electron-multiplying CCDs (EMCCDs) and therefore require significantly higher excitation energy to achieve similar signal-to-noise ratios. The result is rapid photobleaching and a high level of phototoxicity in live-cell imaging, which limits the number of scans that can be acquired before artifacts begin to obscure the experimental results. Line scanning confocal microscopes would be dramatically improved by the introduction of advanced complementary metal oxide semiconductor (CMOS) or EMCCD-type linear charge-coupled device image sensors.
A more advanced approach to high speed confocal imaging involves scanning the specimen with multiple parallel light beams formed by either a rotating Nipkow spinning disk or by scanning a grid array across the specimen to capture all of the image points in parallel. The key element in designing the latter instrument is to create a grid pattern that is capable of scanning the entire field in a single scan motion, yet maintaining the illumination and detection pinholes at large enough separation distances to prevent crosstalk of fluorescence emission between pinholes. The Nikon "LiveScan" swept-field confocal microscope is an excellent example of a multipoint-array scanner that offers the added refinement of selectable grids having different pinhole sizes. The system also enables the alternative option of using a set of slits having different widths to provide maximum excitation energy and emission collection efficiency at the cost of reduced confocality (Figure 3(b)). In operation, a grid of 32 illumination points is swept across the specimen field using a fast galvanometer and piezo element in combination to control the mirrors and capture images at video rate. In slit-scan mode, the swept-field confocal microscope is capable of capturing images at speeds approaching 1200 frames per second. The instrument is also capable of multicolor fluorescence imaging using acousto-optic tunable filter (AOTF) laser control, specialized multi-band dichromatic mirrors, and matched filter sets. In addition, the microscope can be coupled to high-performance, low noise EMCCD camera systems for the most efficient imaging of living specimens using the lowest possible excitation energy, thus minimizing photobleaching and phototoxicity.
Nipkow disk-based spinning disk confocal microscopes offer an excellent alternative for fast scanning of living specimens. These instruments are based on a circular, rotating disk that contains one or more pinhole arrays that are arranged in an Archimedean spiral designed to scan the entire specimen plane during a single disk rotation (or less with some designs). Spinning disk microscopes can also operate with slits rather than pinholes at the cost of decreased axial resolution. Advanced spinning disk systems feature twin disks that contain microlens elements positioned over the pinhole array on the upper disk, which serve to focus the incident excitation illumination on the corresponding pinholes in the lower disk for significantly increased light throughput (illustrated in Figure 3(a)). Instruments of this type are commonly referred to as field, line, or array scanners because they scan an array of illumination spots or lines over the field numerous times during the acquisition of each image. The highest imaging speeds for spinning disk microscopes approach 2,000 frames per second, but the instruments are often hampered in attaining such high rates due to the high rotational speed of the disk necessary to pass each illuminating spot over the same part of the specimen multiple times. The long integration times required by digital cameras used to capture images (at least at full or half frame sizes) also limit the acquisition speed. Spinning disk microscopes are typically illuminated using either lasers, light-emitting diodes, or arc-discharge lamps, but perform optimally when equipped with solid-state lasers of the appropriate wavelength to match fluorophore excitation peaks. For live-cell imaging applications, spinning disk or other array scanning confocal microscopes are often the instrument of choice, especially when rapid cellular dynamics are being investigated. The parallel light beams reduce photobleaching and phototoxicity, enabling longer imaging sessions.
Additional technologies capable of point-scanning the specimen at video rates (although none are commercially available) include instruments equipped with acousto-optic beam deflectors (AODs; Figure 3(d)), digital mirror programmable arrays (DMDs; Figure 4(b)), or rotating polygon-shaped mirrors (Figure 4(a)). AODs utilize ultrasonic wavefronts to generate pressure zones in a crystal that can diffract or deflect incident laser light at an angle that varies with the acoustical frequency. These solid state devices benefit from having few moving mechanical parts and negligible inertia, enabling the generation of highly accurate sawtooth raster scans having almost instantaneous flyback. Furthermore, AOD scanners can produce user-defined deflections to generate scanned regions of interest. The disadvantages of AOD microscopes are based on the dispersive properties of these devices, which are suited only for controlled passage of monochromatic light. As a result, broadband fluorescence emission cannot be descanned and a slit aperture is often used for confocal detection, resulting in a reduction of axial resolution and reduced rejection of fluorescence originating from regions outside the focal plane. In addition, cylindrical lenses are necessary to match the laser beam shape with the rectangular aperture of the crystal. Although these problems have been solved in several experimental instrument designs, the concept of using AOD scanners has not been popular with the microscope manufacturers.
Programmable array microscopes (PAMs) feature a digital micromirror device that acts as a spatial light modulator in the image plane to generate a wide spectrum of user-defined illumination and detection scenarios. These instruments potentially have great versatility in choosing light sources capable of operating in reflection, transmission, and fluorescence mode with optical sectioning and multiple, simultaneous region of interest selection capabilities. Central to the design of PAM instruments is the DMD unit, which consists of miniature (approximately 16 micrometer) square mirror elements that can be individually programmed to rotate either "on" or "off" with respect to incident and reflected light (as illustrated in Figure 4(b)). The DMD enables the use of repeating patterns to increase optical throughput by multiplexing, and they are significantly faster than the galvanometers used in point-scanning confocal microscopes. In principle, PAM instruments can generate images and optical sections that can be viewed through the eyepieces, and fluorescence emission can be directed to a scientific-grade or electron-multiplying CCD camera system. The major advantage of using DMD devices is that they can be controlled by software, have no moving parts (other than the miniature mirrors), and afford the advantage of being able to use the two different mirror positions to direct light through separate optical paths. In this regard, PAM instruments provide more flexibility than spinning disk and slit scanning microscopes. However, DMD microscopes are severely limited on the illumination side. The illuminating beam must be expanded to cover the entire surface of the DMD, reducing the amount of light available from any particular mirror segment, thus rendering the application of laser illumination sources challenging.
Deflection of the light beam with a rotating polygon mirror system is a mature technology that uses optically simple, non-dispersive surfaces to create an output beam having a basic sawtooth raster similar to conventional video scanning. Polygon mirrors (Figure 4(a)) have been utilized in several past commercial and experimental confocal instruments, but are currently not being implemented by the microscope manufacturers. In practice, the use of polygon mirrors requires considerably more complex optical component design, and the units are prone to miniscule variations in reflectivity and angle with respect to the axis of rotation. Known as pyramidal errors, these angular differences produce beam fluctuations that must be optically corrected. Due to the fact that the number of polygon facets must be proportional to the total number of raster scan lines, specialized mirrors must often be fabricated in order to build confocal instruments. Polygons having 15, 25, or 75 sides must rotate at 63,000, 37,800, or 12,600 revolutions per minute, respectively, in order to generate video rate scanning at 15,750 lines per second. These high speeds require specialized bearings, further complicating instrument design. For these reasons, polygon mirror-based confocal microscopes are rare and generally relegated to special interest projects.
Spinning disk, swept-field, and line scanning confocal systems all suffer from compromises in optical sectioning performance when compared to point-scanning confocal instruments. The poorer performance is due to the fact that light from remote regions of the specimen can leak into the detector through spatial crosstalk, and becomes more severe when imaging highly fluorescent thick tissue specimens. In addition, several of the commercially available spinning disk instruments employ a fixed pinhole size that is not adjustable by the user, and therefore these microscopes are limited to being compatible with only specific objective magnifications. The situation is somewhat better for commercially available slit-based spinning disk microscopes that offer interchangeable disks to match objective resolution and magnification. However, most of the instruments described in the paragraphs above sacrifice optimal diffraction-limited illumination and detection for speed, and are unable to produce the pan or zoom functionality available on commercial laser point-scanning confocal microscopes. Nevertheless, these microscopes are capable of generating excellent images of thin, adherent cells in culture at video-rate speeds or better.
Resonant Scanner Basics
Reviewing traditional single-point laser scanning confocal microscopy, the expanded laser beam is scanned over the specimen by using galvanometer mirrors to continuously alter the angle at which light passes through the rear focal plane of the objective. The galvanometers are strategically placed at a 90-degree angle to each other so that the rotational axis of the scan mirrors coincides with their own surfaces as well as the optical axis of the microscope at a telecentric plane (conjugate to the objective rear focal plane). Back and forth rotation of the galvanometer mirrors translates the focused laser spot across the specimen in a raster pattern that traverses the lateral dimensions of the field. The side-to-side rotation of one mirror (x) creates the horizontal scan line while the up and down rotation of the other mirror (y) produces vertical deflection. Modern confocal instruments are equipped with advanced scanning systems that bestow a high degree of flexibility to the shape and size of the specimen region that is scanned. The scan angle can be rotated by re-allocating a portion of the x-galvanometer signal to the y-galvanometer and vice versa. Likewise, the confocal zoom function can be implemented by changing the angular scanning range of the galvanometers. The ability to scan a specialized user-defined region of the specimen is particularly useful for experiments that require photoactivation or photobleaching of selected regions that do not feature a rectangular geometry.
The high-performance servo-controlled, close-loop galvanometers used in laser scanning microscopes are a marvel of engineering, designed to accurately rotate the attached mirrors and position the beam with extreme precision over a lifetime exceeding several billion cycles. Galvanometers are constructed in a manner similar to electric motors using a rotor and a stator to drive the motion of the positioning actuator, which encompasses the entire assembly and responds according to its torque-to-inertia ratio. Depending upon galvanometer design, either the rotor or the stator is a permanent magnetic that interacts with the wire coils of the sister component to produce forces that drive the rotor. Mechanical stops are used to restrict rotor movement to a limited arc of rotation. In most confocal microscopes, moving-magnet actuators are preferred due to their compact size, high positioning accuracy, superior stiffness, and fast response. The galvanometer mirror surface is a thin film of silver or aluminum deposited on a silicon, fused quartz, or beryllium substrate. Among the most important requirements of galvanometer mirrors is that they are lightweight and built of a rigid material, have an aperture size that does not restrict the entrance of light into the objective aperture, and feature a highly polished surface that is flat to a fraction of a wavelength. The mirrors must be equally reflective over a wide wavelength range (approximately 350 to 700 nanometers) and the reflective surface should be centered on the rotation axis of the scan unit.
As discussed above, servo-controlled linear galvanometers are limited in their scanning speed due to inertia and, therefore, can only raster-scan a specimen at image acquisition rates that typically range from 1 to 5 images per second at standard frame sizes. In order to achieve video rates, the slower linear galvanometer used for horizontal line scanning can be replaced with a much faster resonant scanning galvanometer that vibrates at fixed frequency and is the rotational equivalent of a tuning fork. In resonant galvanometers (also known as counter rotation scanners; CRS), the energy stored in a torsion spring or rod assembly is used to oscillate the mirror in a sinusoid manner. These devices cycle at a resonant frequency on the order of 4 to 8 kilohertz, often close to a proportional fraction of the National Television System Committee (NTSC) video standard horizontal frequency of 15,750 lines per second. When a resonant scanner operating at 7,900 Hertz is used to acquire 512 lines that are progressively scanned in a bidirectional orientation (odd lines are left-to-right and even lines are right-to-left), the result is an individual line period of approximately 125 microseconds and image capture rates on the order of 30 frames per second.
Presented in Figure 5 are several resonant galvanometer designs having different mirror apertures (Figure 5(a)), along with a diagram that illustrates the oscillatory motion of the mirror (Figure 5(b)). The high-speed (7.9 kilohertz) resonant galvanometer mirror can rotate approximately 12 degrees in each direction around the central axis. Incident laser excitation light is reflected from the mirrored surface at an angle determined by the physical mirror position (covering a total range of 24 degrees). If the mirror motion is considered in terms of the phase and velocity of the oscillatory cycle, the mirror transitions from a phase of -2π (velocity equal zero) through 0 (maximum velocity) to +2π (velocity equal zero). Thus, as will be described in greater detail below, the velocity is minimal (or zero) at the ends of the oscillatory cycle where the mirror achieves its maximum angle, and maximum in the center of the cycle where the mirror angle is zero.
Among the major advantages of resonant galvanometer scanners is that they gradually accelerate and decelerate as they progress through the oscillatory cycle, thus avoiding the necessity of a reset signal at the beginning of each scan line. In order to achieve smooth motion, advanced scanners are equipped with two oscillating mass-balanced torsion rods that are mechanically tuned to resonate in opposite phase to establish equal, but opposite torques that cancel at the housing attachment position. The result is a reactionless mechanical oscillator that eliminates wobble and external vibration. The mirror aperture size on high-speed resonant scanners is approximately 5 millimeters in diameter, and is fabricated using silver or aluminum coated optical grade beryllium, which has excellent stiffness-to-weight ratio. The scanning angles for these galvanometers range from 15 to 26 degrees, peak-to-peak. During oscillation, as illustrated in Figure 5(b) and Figure 6, the angular velocity of a resonant galvanometer varies in a cosinusoidal manner, reaching maximum velocity at a scanning angle of zero degrees (in the center of the scanning field) and minimum velocity at 90 degrees (at the edges of the field), where the scan direction changes. When generating images using resonant scanners, these non-linear variations present a significant problem in terms of pixel clocking.
The Resonant Scanner Clocking Problem
Due to the non-linearity of resonant scanner velocity, a fluorescent specimen is scanned at the highest speeds in the central region, with the velocity progressively decreasing as the scan reaches the edges. As a result, when the image data flow derived from a resonant scanner is acquired with a frame grabber clocked at a constant pixel rate (which assumes the beam is scanned linearly), the images appear stretched at the edges. Furthermore, the uneven distribution of laser excitation intensity (greater at the edges) produces excessive photobleaching (and potentially phototoxicity) at the edges of the scanned region because of the increased exposure to the laser light. The simplest option for compensating scanning non-linearity is to limit the scanning range to that portion of the oscillation period where the velocity of the galvanometer is almost linear, which occurs over a region spanning approximately 70 percent of the total scan width. Unfortunately, this solution reduces the amount of time that emission signal can be collected, increases the scan turnaround time before signal can be collected again, and doesn't prevent photobleaching in regions of the specimen that fall outside the area being recorded. Photobleaching effects can be minimized, however, when using the linear portion of the galvanometer oscillation by incorporating an adjustable aperture that limits the region of the specimen being exposed to illumination.
Image distortion induced by non-linear resonant galvanometer scanning is fortunately predictable and can be corrected using either software or hardware solutions. Regardless of the correction scheme involved to produce images, the most effective scanning strategy involves collecting data during both the forward and backward scans of the galvanometer mirror. Recording data during the forward scan is straightforward, but due to the fact that the backward scan reverses the direction in which pixels are recorded, the image data must be inverted using specialized read-write buffers or software. In practice, images are gathered at twice the normal width (in effect, 1024 pixels for a final image size of 512 x 512 pixels) and half the normal height (256 pixels). At the end of each forward x-axis sweep of the resonant galvanometer mirror, the vertical sawtooth signal supplied to the linear (y) galvanometer is incremented by one line and the reverse (x) scan commences. In this manner, the vertical scan progresses smoothly for 256 cycles until enough data for one image is acquired.
Both software and hardware clocking solutions for resonant scanning confocal microscopy rely on knowing positional data for the high frequency resonant galvanometer mirror. On 7.9 kilohertz devices, this mirror can rotate approximately 12 degrees in each direction around the central axis (for a total of 24 degrees of rotation). Mirror angular position (θ) and phase or velocity (φ) relationship can be predicted depending upon the state of the oscillating cycle. At the beginning of an oscillation cycle (mirror position equal to plus or minus 12 degrees; see Figures 5 and 6) the mirror is stationary with a velocity equal to zero. As the mirror swings toward the midpoint, the angular velocity increases as a cosine function of the phase and reaches its maximum when the mirror angle equals zero. Continuing, the mirror slows down again as it approaches the end of the forward scan until it reaches a velocity equal to zero as it instantaneously reverses direction. Identical changes in angular velocity are observed as the mirror rotates in the opposite direction. Another factor that should be considered with resonant galvanometers is that these devices are high-Q oscillators and cannot be efficiently synchronized to an external frequency due to large response variations in phase and amplitude when even the slightest drift occurs between the natural galvanometer frequency and the external driving source. Therefore, the resonant galvanometer should itself be used as the master oscillator to which all other timing components are synchronized.
Software-based clocking schemes for resonant scanning confocal microscopy involve algorithms that are capable of pixel data linearization using look-up tables based on the predictable motion of the mirror. As an example, one popular algorithm first determines a phase constant for the mirror and frame grabber, and then locates the center pixel of the final image. This algorithm operates around the central plane of symmetry where the mirror position is known with a high degree of precision. Images are processed by examining one horizontal scan line at a time as they are delivered to the frame grabber. Pixel data is stored as received from the scanner at twice the final image width and half the final height. The first step in the process involves inverting data points from the second half of the horizontal scan line, followed by interlacing the inverted data between the data lines for the first half of the image (Figure 6(b)). Because the midpoint of the image may not be known with certainty due to a lag between collection of data and initiation of the horizontal synchronization signal that triggers acquisition, it is often necessary for the software to offset the inverted scan line by a few pixels. In this case, the right edge of the scan line then serves as a reference point for the image. The next step in the sequence is to apply a correction factor for each pixel or phase position relative to the center pixel. That pixel is then relocated to the correct position in the final image (Figure 6(c)). In the ideal case, using sophisticated high speed computers, incoming data can be analyzed on the fly and recorded to the hard drive in real time.
Because resonant galvanometers can drift with temperature and amplitude variations, as well as being a host to other variables that are difficult to control, the key to designing hardware clocking mechanisms is the precise determination of the instantaneous angular position of the scanning mirrors throughout the oscillatory cycle. The goal is to generate a clock pulse whenever the laser beam traverses an evenly spaced pixel boundary from left to right on the forward scan or from right to left on the reverse scan. These clock pulses will not be evenly spaced in time with resonant galvanometers. As discussed above, this discrepancy occurs because the angular velocity is greatest at the center of the scanning range, where clock pulses will be more frequent (having a spacing of approximately 70 nanoseconds). At the extremes, where the galvanometer mirror begins to change direction, the clock pulses will have a larger spacing (approaching 160 nanoseconds). In the ideal situation, clock pulses are used to trigger analog-to-digital conversion of the photomultiplier output and each pixel will represent an equal spatial displacement in the final images. Several methods have been developed to linearize resonant scanning data (termed scan linearization) using hardware, but the two most successful strategies have been an analog pixel clock and an optical grating to measure mirror position in real time.
One example of the pixel clock solution utilizes a voltage controlled oscillator (VCO) and phase detector to continuously track pixel-to-pixel and cycle-to-cycle operations of the galvanometer. Once the pixel clock is locked onto the scanner resonant frequency, the oscillator clocks a counter, which accesses a memory chip that stores information relating pixel location to scan velocity. The memory sends this stored information through a digital-to-analog converter that feeds the VCO in a closed loop. Control voltages received by the VCO are used to correct the clocking frequency to match changes in the cosinusoidal velocity of the scanner. As a cross-check, the pixel clock sends a signal for each pixel trigger to a phase detector that compares timing to the scanner synchronization electronics. Detected phase errors generate a correction voltage that is fed back to the VCO to maintain cycle-to-cycle synchronization with the scanner. The VCO output is used to clock pixels at the detector. Several advanced proprietary pixel clocks are available from resonant galvanometer manufacturers as options for their control electronics. However, most generate signals that are based on look-up tables and do not allow for small variations in the scanning pattern that occur due to mechanical drift.
The most advanced hardware-based pixel clocks are able to directly track the position of the mirror throughout the scan cycle using optical data obtained from an auxiliary feedback system that directly tracks mirror motion. Perhaps the best example was first described by Roger Tsien and Brian Bacskai and served as the basis for the first commercial confocal microscope containing a resonant scanner, the Nikon RCM-8000, introduced in the early 1990s. A similar clocking system is currently employed by the high-performance Nikon A1R confocal microscope system. In short, the Nikon pixel clock is based on using an auxiliary low-power infrared laser system to reflect a timing beam from the reverse mirror face of the resonant galvanometer. The reflected beam is focused on a grating of clear and opaque stripes (termed a Ronchi grating) and oscillates along the grating in step with the resonant scanner cycle. Light passing through the grating is recorded as intensity variations using a dedicated photodiode that feeds information to the electronic components that digitize the image data. Because the position of the galvanometer mirror is detected with high precision as it oscillates, the pixel clock is extremely accurate.
Illustrated in Figure 7 is a schematic diagram outlining the components of the optical train for a Ronchi grating-based variable pixel clock in a resonant scanning confocal microscope. The primary optical and electronic elements in the pixel clock are shown in the left-hand side of the diagram and include a low-power semiconductor near-infrared laser source, a rotatable disk containing Ronchi gratings of varying spacing, intermediate lenses to focus the reflected laser beam onto the Ronchi grating, and a photodiode sensor that detects light pulses passing through the grating. Components involved in the excitation and imaging optical trains are presented in the right-hand side of the figure, but for simplicity, the number of imaging components is reduced to a minimum. Note that the mirror attached to the resonant galvanometer in Figure 7 has highly reflective surfaces on both sides in order to accommodate the imaging and clocking lasers.
In operation, light emitted by the clocking diode laser is directed to the reflecting backside of the resonant scanner mirror while light from lasers involved in the imaging system are reflected from the frontside of the same mirror. Reflected light from the clocking laser is focused by a condensing lens to sweep across the surface of the Ronchi grating in step with the oscillating motion of the galvanometer mirror. At the grating, the laser beam is scanned across a sequence of alternating opaque lines and transparent spaces of equal width. The condensing lens focuses the clocking beam to a size smaller than the line spacing in the grating to ensure that light is either completely passed or partially blocked by the grating, in an alternating manner dependent upon the position of the galvanometer mirror. A critical point is that the laser beam spot at the grating surface level should be significantly smaller than the grating line width in order to eliminate diffraction. Fluctuations in light intensity induced by the oscillating galvanometer are detected by the photodiode, which is positioned immediately beneath the grating.
During the horizontal scan period of the resonant galvanometer, the clocking photodiode registers pulses of laser light scattered by the Ronchi grating that can be electronically converted to a pixel clock having variable frequency. These light pulses will be more frequent in time when the galvanometer mirror is in the central portion of its oscillatory period, but they will slow down as the mirror reaches the end of a scan and reverses direction. Among the requirements of the photodiode detector are the ability to handle the variable pulse frequency (in terms of bandwidth) and it must be large enough to detect the entire line sweep. The detector is connected to a transconductance amplifier that transforms photodiode current pulses into an amplified voltage, which is then fed to a frequency doubler circuit that doubles the number of pixels per line. The photodiode and associated electronics deliver 256 pulses per line and the frequency doubling circuit transforms this output into 512 pulses per line that are acquired by a frame grabber and used to construct an image.
Presented in Figure 8 is an illustration of the basic concepts behind the use of a Ronchi grating coupled to a photodiode detector as a variable pixel clock. For purposes of clarity, the grating illustrated in Figure 8(a) contains only eight spacings of equal width superimposed with one complete cycle of cosinusoidal motion from the resonant galvanometer versus time (red curve; 125 microseconds). The actual grating has 256 line spacings and is far more compact in the linear dimensions due to the fact that the pixel clock laser actually retraces a single line (perpendicular to the opaque lines) across the grating surface in operation (Figure 8(b)). As the light beam enters and exits the clear regions of the Ronchi grating, the transmitted light intensity that is detected by the photodiode rises and falls in step with galvanometer motion (Figure 8(c)). Each transition (in this case, from dark to light) of light intensity detected by the photodiode is used to generate the pixel clock (Figure 8(d)) that triggers image capture by the microscope. Although the clock pulses illustrated in Figure 8(d) are non-uniformly spaced in time (ranging from approximately 70 to 160 nanoseconds), they nevertheless correspond to equal intervals in image space.
In order to acquire pixel samples at equal spatial intervals (rather than in equal temporal increments), photomultiplier signals from the primary microscope imaging system pass through conventional amplifiers with variable gain and offset, and are then digitized by an analog-to-digital converter triggered by the pixel clock generated from the Ronchi grating pulses. However, half of the image information is inverted due to the fact that the resonant galvanometer sweeps in both directions. In order to rectify this bidirectional scan artifact before constructing an image, the digital pixel output is first transferred to either a first-in-first-out (FIFO) or last-in-first-out (LIFO), buffer depending upon whether the pixels were acquired from left to right or right to left. Thus, during the horizontal sweep of the galvanometer mirror, image pixels that are acquired during the first half of the sweep (time period of 62.5 microseconds) are fed to the FIFO buffer whereas pixels acquired during the reverse sweep are fed to the LIFO buffer. The FIFO buffer is read and reset to zero at the beginning of each horizontal line acquisition, whereas the LIFO buffer counts up during storage of incoming data and down during the subsequent readout to ensure that pixels acquired during the reverse sweep are inverted.
Advantages of Resonant Scanning
Unlike the situation for spinning disk, swept-field, and rotating polygon-based microscopes, resonant scanning laser confocal microscopes are able to alter the magnification without changing objectives by using the versatile confocal zoom functionality. The physical basis for zoom control involves altering the rotation angle for both the horizontal resonant galvanometer scanning mirror as well as the vertical linear galvanometer mirror. The smaller sweep angle reduces the area of the specimen being scanned while maintaining the same number of pixels to effectively provide an enlarged image of selected specimen details. Invoking the confocal zoom function requires changes to the spacing interval of the lines in the Ronchi grating used for the variable pixel clock. The disk in Figure 7 contains Ronchi gratings of four sizes, each of which can be rotated into the pixel clock optical train in order to change the zoom factor. When the line size and spacing dimensions in the Ronchi grating are reduced by half (maintaining the same total number of lines), the zoom factor increases by a factor of two. Likewise, reducing the grating line size dimensions by a factor of four increases the zoom factor to 4x. The number of possible zoom settings depends upon the number of Ronchi grating sizes available to the variable pixel clock. The Nikon A1R laser confocal microscope is capable of 7 zoom steps ranging from 1x to 8x. Interchanging gratings requires reducing the sweep angle of the resonant galvanometer, and this is accomplished using pre-stored information in the microscope control electronics that coordinates galvanometer sweep angles with the selected grating.
Resonant scanning galvanometer dynamics provide wide latitude in choices for imaging acquisition rates simply by altering the dimensions of the acquired images, much the same as in pixel binning by the area array detector in spinning disk microscopes. Although the scan rate of the horizontal line is fixed by the resonant period of the galvanometer, the number of lines used per image can be reduced to generate faster imaging rates. Thus, at an image size of 512 x 512 pixels, the Nikon A1R microscope produces images at the rate of 30 frames per second. If half the number of vertical lines are collected to acquire an image having dimensions of 512 x 256 pixels, the frame rate is theoretically doubled to 60 frames per second. Note that the image size-speed relationship is not exactly linear due to slight delays in repositioning of the linear (y) galvanometer at the higher speeds. For the Nikon A1R microscope, the highest scan rate achievable in resonant scanning mode is 420 frames per second at an image size of 512 x 32 pixels. For many applications, such as fast calcium wave imaging in live cells, the reduced vertical image dimension is an acceptable sacrifice for the increased speed.
Illustrated in Figure 9 are photoconversion and photoactivation experiments using optical highlighter fluorescent proteins in living cells, which significantly benefit from a combination of the tandem scanning and high speed capabilities of so-equipped resonant scanning confocal microscopes. Figures 9(a) to 9(c) demonstrates photoconversion in gap junctions labeled with a fusion of mEos2 (a red-to-green photoconvertable fluorescent protein) to the connexin 43 gap junction component and expressed in live HeLa cells. A small region is selected for photoconversion with the linear galvanometer coupled to a 405-nanometer laser (Figure 9(a)), followed by imaging with a 488-nanometer laser (Figures 9(b) and 9(c)). Figures 9(d) to 9(f) show photoactivation of PA-GFP fused to human beta-actin in kidney cells. Similar to Figure 9(a), the region of interest is illuminated with a 405-nanometer laser followed by imaging at 488 nanometers. Finally, photoswitching of mitochondria using a fusion of red-emitting kindling fluorescent protein (KFP1) to a mitochondrial targeting signal is presented in Figures 9(g) to 9(i). Labeled mitochondria are imaged with a 561-nanometer laser in both fluorescence and differential interference contrast (DIC) mode in Figure 9(g). After completely photoswitching the labeled chimera "off" with 488-nanometer illumination, the mitochondria appear devoid of fluorescence (Figure 9(h)), which can be reactivated (Figure 9(i)) by again imaging at 561 nanometers. Investigations using optical highlighters can be conducted with resonant scanning confocal microscopy at imaging speeds necessary to elucidate intracellular dynamics on a wide variety of timescales.
These imaging techniques are made possible by the hyperselector in the Nikon A1R scanner (see Figure 1), which diverts the 405-nanometer photons used for photoactivation to the galvanometer scanner, leaving the resonant scanner free to simultaneously acquire high speed images of the photoactivation products. The galvanometer scanner can be zoomed to restrict the size of the stimulation area (which can dramatically accelerate photoactivation). The ability to photoactivate at 405 nanometers while simultaneously acquiring images using the resonant scanner extends to the Nikon A1R-MP multiphoton microscope.
Traditional laser scanning confocal microscopes suffer from the major disadvantage of being relatively slow with regards to image acquisition speed despite their high resolution and superiority in the rejection of light that arises from fluorescent regions removed from the focal plane. The primary technical limitation in confocal technology is the requirement for a fast horizontal (x-axis) scan mirror that is capable of producing 500 or more lines per frame in real time. However, even at the rather slow scanning rates of those point-scanning confocal microscopes equipped with standard linear galvanometers, the beam dwell time that constitutes a single pixel is very short and limits the amount of signal that can be collected. As new technologies are introduced that increase confocal scanning speeds, such as the resonant galvanometer, the argument persists that continuing to increase point-scanning confocal microscope speeds will only result in decreasing the amount of signal until it eventually runs out at the highest speeds. This overriding, but mission-critical signal-to-noise concept has perhaps unfairly led to the development of several alternative high-speed imaging techniques that deliver inferior optical sectioning performance at the expense of focusing engineering efforts on point-scanning instruments capable of video rate image capture.
Contrary to expectations, several reports have emerged that high-speed point-scanning confocal microscopes equipped with resonant scanners may be capable of both improved signal levels and reduced photobleaching over what had been anticipated. This unexpected result most likely arises from the exceedingly complex photophysical behavior of synthetic fluorophores, quantum dots, and genetically-encoded fluorescent proteins, all of which appear to be capable of photoactivation, photoconversion, and photoswitching to various extents through a wide spectrum of as yet unexplained mechanisms. In regards to resonant scanning confocal microscopy, the increased fluorescence signal may arise from an excitation illumination dose-dependent response that is modulated by the rapid scanning to yield increased signal levels. Typically, such a response should depend on the diameter of the focused laser spot and the speed at which the beam scans the specimen. The result mimics pulsed illumination of the fluorophores. Because pulsed illumination may transition fluorophores into a dark or "off" photoswitched state in which they avoid entering the triplet state, the level of fluorescence emission is increased while photobleaching rates are somewhat reduced. These reports have not been experimentally verified on a widespread basis, but they should be borne in mind when examining the results of fast confocal imaging of living and fixed specimens.
Resonant scanning confocal microscopy, a technology that is still being developed and refined, should prove to be a useful method for examining a host of biological issues using cells, tissues, and complete organisms under conditions where video rate imaging is critical to the experiment. The ability to simultaneously utilize resonant and linear galvanometers to conduct high-speed imaging in tandem with specific region-of-interest photoactivation or photobleaching at superior optical section resolution is an added benefit that cannot readily be achieved with other techniques. The most sophisticated instruments are able to combine video rate confocal scanning with spectral imaging using specialized detectors to further enhance the utility of these microscopes. For example, imaging calcium waves using fluorescent protein FRET biosensors requires very fast acquisition speeds because the wave can traverse the cell in less than a second. Using traditional confocal microscopes, calcium imaging is virtually impossible, but with an instrument such as the Nikon A1R, the wave can be captured in real time and the contributions of both fluorophores to the FRET emission spectrum can be unraveled with the spectral detector. This type of experiment, as well as countless others, will benefit in the future from examination with resonant scanning laser confocal microscopes.
Jeffrey M. Larson and Stanley A. Schwartz - Nikon Instruments, Inc., 1300 Walt Whitman Road, Melville, New York, 11747.
Michael W. Davidson - National High Magnetic Field Laboratory, 1800 East Paul Dirac Dr., The Florida State University, Tallahassee, Florida, 32310.