SEARCH

SEARCH BY CITATION

Keywords:

  • confocal microscope;
  • laser power stability;
  • quality assurance;
  • spectroscopy;
  • slide-based cytometry

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. SUMMARY
  6. Acknowledgements
  7. LITERATURE CITED

Background:

All slide-based fluorescence cytometry detections systems basically include an excitation light source, intermediate optics, and a detection device (CCD or PMT). Occasionally, this equipment becomes unstable, generating unreliable and inferior data.

Methods:

A number of tests have been devised to evaluate equipment performance and instability. The following four instability tests are described: galvanometer scanning, stage drift, correct wavelength spectral detection, and long-term laser power.

Results:

Quality assurance tests revealed that a confocal microscope can become unstable in the following parameters, yielding inaccurate data: laser power, PMTs functionality, spectrophotometer accuracy, galvanometer scanning and laser stability, and stage drift. Long-term laser power stability has been observed to vary greatly.

Conclusions:

Confocal systems can become unstable in the following parameters: long-term laser power, galvanometer scanning, spectrophotometer accuracy, and stage stability. Instability in any of these parameters will affect image quality. Laser power fluctuations result from either a defective Acousto-optic tunable filter or improper heat dissipation. Spectrophotometer instability will generate unreliable spectra data, extra light reflections, and poor image quality. Galvanometer scanning instability yields poor image quality while microscope stage drift results in a sample going out of the plane of focus. With minor modifications, these tests may be applicable to other slide-based systems. © 2006 International Society for Analytical Cytology

Slide-based cytometry microscope systems are useful to obtain intensity measurements from biological samples (1–4). Slide-based cytometry detections systems encompass a number of different optical configurations. Although the configuration of the system may appear to be unique, the basic optical principles used are similar between all of the optical equipment that is currently used for slide-based analysis. All of these systems basically include an excitation light source which may be laser, xenon, or mercury arc, intermediate optics which may include an Acousto-optic tunable filter (AOTF) dichroics, barrier filters, and lenses, and finally a detection device which may be a CCD camera or a PMT. Scientists should be cognizant of the fact that all optical systems can become unstable during their normal operation even thought they pass an initial quality assurance test.

The confocal laser-scanning microscope (CLSM) usually consists of the following: a standard high end microscope with very good objectives, different wavelength lasers to excite the sample, fiber optics to deliver the laser light to the stage, AOTF to regulate the laser light delivered to the stage, barrier filters, pinholes to control eliminate the out of focus light, electronic scanning devices (galvanometers) to acquire the image, and detection devices to record the number of photons (i.e., PMTs). Various other electronic components are included to control these devices. For this system to operate correctly, it is important for it to be properly aligned and to have all the components function correctly. Instrument performance tests included the following: laser power, laser stability, field illumination, colocalization, spectral registration, spectral reproducibility, lateral resolution, axial (Z) resolution, lens cleanliness, lens characteristics, and Z-drive reproducibility (5–17). This list is not inclusive and additional parameters may be needed to assess CLSM performance.

During the quality assurance testing procedure of a confocal microscope, it was found that some of these tests acquired data appeared to fluctuate over time. Many of these tests indicated that the system was delivering unreliable data that will result in bad image quality and needed a service call. The lack of adequate power stability influenced the accuracy of fluorescence measurements. (5, 6, and 10). These confocal systems have lasers that usually have less than 1–2% long-term (hours) power fluctuations. However, confocal systems usually exceeded 10% laser power fluctuations (5, 6). Occasionally, as described later in this communication, our laboratory has observed power fluctuations in excess of 400%. The lack of spectral reproducibility in this system made this data endpoint unreliable (8, 9) and suggested a condition of general instability.

The parameters described in this communication are particularly troublesome as the machine was checked out with many of the standard tests previously described before running the experiment and it yielded satisfactory results. During the acquisition of image data, somehow the system became unstable. The acquisition of instability data on any scientific equipment is difficult to study because of its intermittent nature and the lack of understanding the conditions necessary to generate the data. Tests must be devised to monitor instability. This manuscript will describe instability in four parameters: laser power, galvanometer scanning, stage stability, and spectroscopic stability. In all four cases, the machine was evaluated to show how data can unknowingly be altered during acquisition. It is hoped that this type of publication will make both the scientific community aware of potential problems that may affect their data and make the manufacturers aware of seriously problems that may exist on their equipment which should be addressed to make them more stable and reliable. Many of these problems could be eliminated if there were published specifications on the equipment and approved tests to measure these specifications. Without the use of specifics tests and known performance specifications provided by the manufacturers, the scientists and manufacturer may not even be aware of instability on their equipment.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. SUMMARY
  6. Acknowledgements
  7. LITERATURE CITED

The methods described are primarily for a Leica SP1 confocal microscope and are represented as individual steps to facilitate being adapted. Relevant comments and discussion are made in each described section to facilitate the understanding and adaptation of the procedure to other slide-based equipment.

Confocal Microscope

The majority of data presented in this manuscript was derived on a Leica TCS-SP1 (Heidelberg, Germany) confocal microscope system. This system contained an 75 mW argon–krypton gas laser (model 643, Melles Griot, Omnichrome) emitting 488, 568, and 647 nm lines and a 60 mW Coherent Enterprise UV laser emitting 351 and 365 nm lines. The power of the laser lines can be measured at various points in the optical pathway. There was a considerable amount of laser light being suppressed by the optical components of the light path, by the time the laser was measured on the stage it was usually only 1 mw for visible laser lines and 0.5 mw for UV laser line. The Leica SP1 system contains an AOTF, with the following three dichroics for visible light applications: single dichroic (Reflection shortpass (RSP 500); double dichroic; and triple dichroic and a RT30/70 beamsplitter.

These tests derived on a Leica SP1 system were shown to be applicable to other point scanning systems that contain different types of lasers, objectives, or other hardware configurations. A Zeiss LSM 510 unit containing three lasers {argon 488 (25 mw); HeNe 543 (1 mw); and HeNe 633 (5 mw)} with a merge module and an AOTF was used. A Leica AOBS unit that contained four lasers UV (Argon 365 Enterprise 60 mW) {Argon 488 (50 mW); HeNe 543 (1.2 mW); and HeNe 633 (10 mW)} with a merge module and an AOTF was also used.

Beads

Fluorescence microscope test slide # 2 (FocalCheck™ beads; F36913, Invitrogen) was used in this study. The eight beads on the slide have the following excitation/emission ranges: Green 503/511, 511/524; Orange 541/555, 545/565; Red 578/605, 589/613; Far Red 657/676, 671/692. This slide contained two rows of five wells. One row contained two different beads with spectra that differed by only 10–15 nm. The second row consisted of these two dyes that are contained on the same bead as a core and a ring configuration. The fifth well of each row contained a mixture of either the four ring beads or the eight different spectral beads. This slide could be used for spectral unmixing studies and spectral registration studies.

Biological Test slides

FluoCells® prepared slide #1 (F-14780, Molecular Probes Eugene, OR) used as biological test slide. FluoCells prepared slide #1 contains bovine pulmonary artery endothelial cells. The mitochondria are stained with MitoTracker Red CMXRos, F-actin is labeled with BODIPY FL phallacidin, and the nuclei are labeled with DAPI.

FluoCells® prepared slide #6 (Molecular Probes, F36925) were Muntjac cells stained with mouse anti-OxPhos Complex V inhibitor protein, Alexa Fluor® 555 goat anti-mouse IgG, Alexa Fluor® 488 phalloidin, TO-PRO®-3 staining nuclei.

Fluorescent plastic slides.

The red Chroma slide (Chroma, Battleboro, VT) was used for excitation at all wavelengths including UV. Occasionally, the blue Chroma slide was used for UV excitation.

Square Sampling and Galvanometer Check

It is important to ascertain whether there was square sampling or rectangular sampling in an image. Leica uses a computer chip that is glued onto a glass slide as a test substrate. A commercial product can be obtained from MicroBrightField Scientific (Williston, VT), Geller (Topsfield, MA), or Richardson Technologies (Toronto, Canada) that has square boxes with vertical lines. A digital TIFF image was obtained using a dry 20× objective.

PMT Spectral Evaluation

The PMT spectral response was measured over a large spectral region using an inexpensive fluorescence calibration lamp called a (multi-ion discharge lamp [MIDL]). It consists of a defined mixed ion gas (816025, LightForm, Hillsborough, NJ), which has been previously described (8–10). Spectra beads (6u) can be used to evaluate the spectral reliability of the system. However, they are not as accurate as the MIDL lamp in detecting problems but may show gross changes in PMT spectral.

Spectral lambda scan.

The spectral scan is made using a single reflecting mirror obtained from Leica (Leica Microscystems, Mannheim, Germany) or Edmonds Scientific, (Barrington, NJ). A 21 mm square (#31008 Edmonds Scientific, Philadelphia, PA) was glued onto a microscope slide and a cover glass (#1.5 Fischer, Pittsburgh, PA) was placed on top of the slide with a drop of immersion oil (Leica Immersion oil, n = 1.518). The cover slip is placed firmly onto the mirror to remove all excessive oil. This type of standard test slide can also be obtained from a confocal manufacturer or Spherotech (Libertyville, IL) (1, 2).

Software Analysis

The analysis of the images was made on workstations that contained Leica or Zeiss software packages. Imaris (Bitplane, Zurich, Switzerland) was used for 3D visualization of data. The TIFF images were also imported into Image Pro Plus (Media Cybernetics, Silver Springs, MD) or ImageJ (NIH) for more intensive measurements and analysis.

RESULTS AND DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. SUMMARY
  6. Acknowledgements
  7. LITERATURE CITED

Laser Stability (Visible)

Laser stability measurements are essential for requiring intensity measurements or time dependent physiological experiments. In our system, there was periodic noise in the laser power tested by the transmitted light detectors that exceeds the manufacturers specifications (Melles Giriot) of laser stability fluctuations of less that 0.5% over a 2-h time period. The laser lines had a periodic cycle that varied at different times as shown in Figures 1 and 3 (5). Although argon lasers and helium neon lasers (543 nm or 633 nm) are considerably quieter, it is believed that most of the intensity fluctuations in the confocal system are being derived from one of the components in the excitation light path and not the argon krypton laser itself.

thumbnail image

Figure 1. Visible laser stability. The laser power fluctuations was measured with a 10× objective using a Chroma red slide excited with a 488 and 568 lasers (which is dark and which is light). The fluorescence was measured sequentially 400 times every 30 s for total time duration of 3.33 h. Similar data were obtained using transmission detector without slide, a mirror slide, and 10u Spherotech beads. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Excitation power can be measured at various points in the optical path. These include intermediate positions located directly after the laser, after the fiber optic, after the AOTF, and at the stage. The emission power can be measured with fluorescence or transmission PMT. The tools to measure power can be just a simple power meter with the appropriate laser detector positioned on the stage or at other places in the optical path. The readout device can be a chart recorder, computer, or paper and pen. On the emission side, we can measure the mean of a Region of Interest (ROI) and observe how this value changes over time. Although the cleanest way to measure the power is with a transmission detector without a slide in the optical path, a similar test can be made with a mirror, a Chroma fluorescence slide, or fluorescence beads. However, by using slides in the optical path, the stage becomes a factor that can influence the data and must be considered.

In a confocal microscope, there are different ways to measure power stability over time (hours). These include the following: (1) manufacturers installed pin diodes; (2) laser meters on the microscope stage connected to a readout device; (3) fluorescent emission intensity from a plastic slide detected by a PMT; (4) transmission optical system detection; and (5) XZ or XY reflection of a mirror slide. At the time we tested the equipment, the pin diode test was not stable and should only be used as a subjective assessment of power. In our current system, Leica SP1 system, these diodes are not functional.

Laser power measurements were made continuously for a few hours to evaluate stability and possible fluctuations in power using a few different configurations that yielded similar results (5, 6, and 10). A fluorescent plastic slide was placed on the stage to measure laser power stability using in PMT 1 (blue light sensitive, low noise, R6857) and also in the transmission detector (Hammatsu, R6350). The laser power needs to be reduced with either neutral density filter or with the AOTF adjustments to eliminate possible laser bleaching of the slide. The transmission detector (R6350) can also be used without a slide in the light path possible eliminating stage problems. This configuration eliminated any possibility of bleaching or laser interaction with the substrate and was considered to be the superior way to measure laser power. A reflecting mirror slide can also be used in place of the fluorescent slide. In favor of the mirror slide or a Chroma fluorescent slide is that they allow the investigator to use the same PMTs that will be used to detect fluorescent signals from biological samples. In such cases, extreme care should be taken to lower the laser power with an AOTF to reduce possible interaction of the laser beam on the sample. Sometimes, the output fluorescence intensity may increase because of repetitive excitation or it may decrease because of bleaching.

  • 1
    To measure laser stability using the transmission optical system (R6350 detector), the microscope is first aligned for Kohler illumination with a histological slide, which is then removed from the optical path.
  • 2
    The laser power is adjusted by a combination of laser power and AOTF adjustment of the laser line power. It is desirable to adjust the AOTF so that the transmission detector voltage remains constant for all three wavelengths. Since the transmission detector is more sensitive in the blue range, it is best to adjust the AOTF for 647 nm (or 633 nm) first and then 568 mm (or 543 nm) and finally 488 nm laser line.
  • 3
    The image intensity is measured sequentially using the one transmission detector for the three wavelengths of the argon–krypton laser (488 nm, 568 nm, and 647 nm).
  • 4
    The test usually takes a few hundred scans separated by 10–30 s over a period of 2–3 h.
  • 5
    To plot the data, a large ROI of the three excitation field images are determined and the means are plotted over time using software provided by the manufacturer as shown in Figures 1–4. The goal of this test is to have a straight line with no variations in power. It is not necessary to save the images in the scan but it is useful to save the data measurements in an Excel file, text file, or equivalent file format.
  • 6
    The goal of this laser power stability test is a flat stable line. However, an example of the type of laser power stability data that can be achieved with this test is represented in Figure 1. There is periodic noise in the laser system that exceeds the manufacturers (Ominichrome) laser stability fluctuations specifications of less that 0.5% over a 2-h time period. The 488 nm and 568 nm laser lines have a periodic cycle and stability are never achieved (Fig. 1). The 647 line also shows periodic fluctuations but is not represented in Figure 1 for clarity of the figure.
  • 7
    To use the mirror slide, it is placed on the stage, the sliders are put over each laser line, and the PMT voltage adjusted to maximum intensity without saturation. Data from all three laser lines can be obtained simultaneously every 10–60 s.
  • 8
    The Chroma red slide can be used to measure stability in PMT 1 for the 365, 488, and 568 nm laser lines. However, if all three are measured simultaneously, it has to be used in sequential scanning mode with very low laser power to reduce bleaching. It is advantageous to use the same PMT at the same voltage setting. The AOTF should be used to regulate the laser throughput power in the system. Minor modifications to this procedure will have to be made in dichroic-barrier based systems not containing the slider configurations of the Leica system.
thumbnail image

Figure 2. UV laser stability of a coherent enterprise UV laser. The coherent enterprise UV laser delivers less than 1% peak-to-peak noise. The laser was connected to an LP 20 water–water cooler, which should be set at least 10°C above the cooling water of the building. It should also be set above the ambient temperature of the room. Improper set points for laser cooling resulted in bad thermal regulation of the laser (B). Improper fiber alignment resulted in additional laser intensity variations (C). The elimination of the temperature and polarization issues resulted in proper laser stability (A, 3% power variation over time). The test was conducted by measuring the laser power stability in PMT 1 using a blue fluorescent plastic slide. Neutral density filters were used to reduce the power and thus minimize slide bleaching. The transmission detector optics was also used to measure UV laser stability and it gave similar results to using a PMT with a UV excitable fluorescent plastic slide.

Download figure to PowerPoint

thumbnail image

Figure 3. Visible laser stability. The laser power fluctuations using a 488 nm and 365 nm lasers were determined using a 10× objective and a Chroma red slide. The fluorescence was sequentially measured every 30 s (400 times) for total time duration of 3.33 h. Similar data were obtained using transmission detector without slide, a mirror slide, and 10 μm Spherotech beads. The variation of the peak to peak using 488 nm excitation is five fold or approximately 500%. The UV straight line suggests that the system scanning and detection are working correctly but the instability is located in the visible light path. An unstable AOTF may be contributing to this 488 nm sinusoidal pattern. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

thumbnail image

Figure 4. Visible laser stability after AOTF replacement. The laser power fluctuations of the 488 nm (middle line), 568 nm (bottom line), and 647 nm (top line) were measured with a 10× objective using a mirror in a warm system (on for 1 h). The fluorescence was measured sequentially 800 times every 15 s for total time duration of 3.33 h. The scans show reliable stability with variations in the order of less than 10% for each line. The 488 [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Laser Stability: Temperature Effects and Heat Dissipation

The Coherent UV Enterprise laser (365 nm) delivers less than 1% peak-to-peak noise and is considered to be a very stable laser. It is an argon gas laser that is water cooled. The Coherent laser was tested in a similar manner to that described for visible lasers using either the transmitted optical system or the PMT system with blue colored fluorescent plastic slides. Very low laser power was used. A relatively stable line showing minimal fluctuations should be obtained (Fig. 2A). One source of power instability appears to be way the gas laser is cooled and how the laser heat is dissipated from the system. If the temperature of the cooling system is not regulated properly, power fluctuations may occur. This was illustrated with our Coherent Enterprise UV laser that was connected to a Coherent LP 20 water–water exchange cooler. This cooler should be set at least 10°C above the circulating cooling water of the building and it should be set above the ambient temperature of the room. Improper set points for the LP 20 cooler resulted in temperature regulation problems of the circulating cooling water in the laser, which in turn resulted in the improper regulation of the laser power (Figs. 2B and 2C).

Similar type of power fluctuations occurred in the visible system that had improper heat dissipation (data not shown). The dissipation of heat is a very important variable to consider in measuring laser stability. Improper heat dissipation with the air-cooled lasers will result in laser power fluctuations. In our case, we have observed fluctuations with an argon krypton air-cooled laser that (1) had a restrictor in the exhaust line and (2) used a smaller exhaust duct (4″ instead of 5″) to remove heat. In both cases, the heat was not dissipated correctly and the laser power in the CLSM fluctuated above 20%. Removal of these heat dissipation problems appeared to reduce the laser fluctuations to less than 10% (5). All lasers have to dissipate heat properly and their thermo regulators must be set correctly or the laser power will fluctuate as illustrated using the Coherent Enterprise laser in Figure 2. It is useful to attach an inexpensive indoor–outdoor thermometer in the exhaust system to determine laser exhaust temperature and room temperature. A rise in temperature signifies improper temperature regulation and possible laser instability. Room temperature should be kept cool and at a constant temperature.

Power Instability—AOTF

The CLSM systems of the 1990s used manual laser power and neutral density filters to regulate the intensity of laser light that illuminated the samples. Currently, the AOTF is very popular as laser excitation can be very accurately controlled and only the Zeiss Pascal does not have an AOTF. The AOTF is enormously useful in operating a confocal microscope as wavelength selection and power can be very easily and accurately controlled. This effectively acts to reduce the cross talk between the excitation wavelengths. However, the AOTF may be introducing power fluctuations in the CLSM system by improper thermal regulation or improper RF frequency modulation. The AOTF is a birefringence crystal capable of rapid and precise wavelength selection. The exact precision and accuracy of the unit in the confocal microscope has not been described with the specifications of the system. Apparently, the original AOTF that was installed in older CLSM models were not temperature regulated and may have introduced power instability in confocal microscope systems. This stability problem is very disturbing to investigators expecting to make comparative intensity measurements on biological samples. One needs to understand how much power fluctuations are occurring in a system prior to making intensity measurements (5).

To illustrate this power fluctuations, power intensity was measured sequentially from the UV laser and the visible laser. Figure 3 shows the fluctuations of the 488 nm excitation from the argon–krypton laser and the UV excitation derived from the Coherent Enterprise laser. This test was made by placing a Red Chroma slide on the stage and sequentially measuring the 488 emission and then the UV emission using the same PMT settings and detection range over a 3.3 h time period. Care was taken to reduce the laser power to decrease bleaching the Red Chroma slide. The data show that the UV line is stable, while the visible line shows fluctuations in excess of 400%. This suggests that the scanning mechanism and the detection mechanism are properly functioning but a problem exists with the visible excitation light path or one of its components in this light path. Since the scanning mechanism and detection optics are the same after the sample excitation, it suggests that there must be something in the excitation light path that is causing this problem. Possible components in the visible light path include the following: laser, fiber optics cables, AOTF crystal, or one of the electronic boards controlling this crystal.

After these data were obtained over a period of almost 7 years, the AOTF was replaced and there was an immediate improvement of laser stability to approximately 10% or less over a 2-h time period (Fig. 4). The periodic fluctuations previously observed were eliminated. It is highly suggestive from this observation that the AOTF may be one of the major causes of instability as the laser power fluctuations were eliminated by the simple replacement and proper alignment of the AOTF.

Power Stability: Fiber Optic and Polarization

Laser stability tests over time may be related to a mismatch between the laser polarization and fiber optic polarization. The fiber optics may introduce power fluctuations because of the deterioration or breakage over time. Fiber optical problems will attenuate the laser power which necessitates using higher laser power or higher PMT settings in the operation of the CLSM.

  • 1
    One test procedure recommended by Leica to ensure that polarization is correct after the alignment procedures is to wiggle the fiber optic and see if the image returns to the same intensity values, suggesting that the polarization is correct. This is a fairly crude test but it will demonstrate whether the system fiber optic needs further polarization alignment. Better procedures are being tested by all the manufactures to eliminate this fiber optic–laser polarization alignment problem.
  • 2
    A second procedure recommended by Leica technicians is to observe a 1u bead in interference contrast with the microscope and then observe it with the confocal. The bead should show a line diagonally down the center that stays stable over time.
Square pixels, phase alignment, galvanometer test.

The number of small boxes observed was counted by eye in the vertical and horizontal directions. If there is the same number of boxes per inch in the vertical and horizontal directions, then it can be assumed that the sampling of pixels is square. A more accurate measurement can be made by measuring the length and width of the boxes. If the vertical and horizontal lengths are not equivalent, then rectangular boxes will occur, which is undesirable for accurate pixels and imaging. Using this test grid, the vertical lines also serve as an indicator to determine if the galvanometer is scanning correctly. If the scanning length in the X direction is not the same as the scanning length in the Y then the pixels will not yield a perfect square and the measurements of objects will be distorted. This value should stay constant, but it must be checked occasionally and adjusted, especially if a galvanometer is replaced.

Bidirectional scanning is useful to reduce the length of time it takes to scan an image. The phase adjustment in bidirectional scanning must be checked and adjusted to ensure that the scan in both directions line up correctly. A lack of adequate phase alignment will result in a decrease in resolution by mismatching the scans in the left direction from those in the right (opposite) direction. In cases that require the highest resolution image, unidirectional scanning should be made even though it will take twice as long to acquire the image.

The pixel size and symmetry in XY directional field scanning can be checked by using a computer chip attached to a glass slide (Leica) or a slide obtained from MicroBrightField Scientific, Geller MicroScientific, or Richardson Scientific (Fig. 5A).

  • 1
    The confocal should be set up in reflective mode (RT 30/70) using a 10× or 20× dry lens.
  • 2
    The small boxes or reticule lines in the vertical and horizontal direction should be compared by counting or by a measured standard line. This test ensures that the scanning in the X and Y directions yields a perfect square and the information will be registered correctly. If there is the same number of boxes per inch in the vertical and horizontal directions, then it can be assumed that the sampling of pixels is a square. If they are not equivalent, then the sampling of pixels will be rectangular, which is not desired.
  • 3
    This test should be done with slide-based system to ensure that the magnification is correct.
  • 4
    If unequal scanning pattern exists (Fig. 5A), it suggests a galvanometer problem. A galvanometer problem is also shown by the jagged edges of the vertical lines. The insert in Figure 5A represents the correct pattern.
thumbnail image

Figure 5. A MicroBrightField slide with a rectilinear grid can be used to detect that the scanning field is actually square and not rectangular using with transmission optics or reflection mode with the detector PMT placed directly over the laser line. (A) Boxes that are not square, suggestive of a malfunctioning galvanometer on an AOBS system. The jagged edges on the vertical lines in the scan also suggest a galvanometer problem. If a horizontal line is drawn over a few of the squares (B) the width should be equivalent for one image or 20 images that are sequentially averaged. Any widening of the line will indicate a badly functioning galvanometer. Tests should be run with a zoom setting of 1 and 20× lens. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

A test substrate that has horizontal and vertical lines should be used to check the galvanometer. Twenty sequential images of a straight line (i.e., MicroBrightField slide micrometer) should be obtained and compared with a single image of the straight line (Fig. 5B). The instability of the galvanometer is shown by (1) the line widening, (2) the existence of jagged edges, and (3) unequal scanning rectangles (Figs. 5A and 5B). The line should not change in thickness and the line should be entirely straight. Any shifting of the line thickness suggests a problem exists with the galvanometer. Sometimes jagged edges can occur as shown in Figure 5A, which also suggests a galvanometer problem. The test can also be done with other types of uniform objects or beads. Leica uses a 1 μm bead observed with interference optics as a slit in the middle. If there is movement of the edges or jumping of the bead, it suggests a problem with the scanning mechanism.

Stage Instability

If the stage is mechanically unstable, the sample will go out of the plane of focus. In our current system, it took over 60 min for the electronics to stabilize and to eliminate microscope stage drift. A mirror slide can be placed on the stage and a XY or XZ scan can be sequentially acquired over time as described in the power stability section. The same test can be made with a Chroma red slide by measuring intensity of an image sequentially acquired over time. When using the Chroma slide, it is critical to reduce the laser power with an AOTF or neutral density filters to minimize the possible bleaching of the slide. The transmission optics can be used as a comparison to make sure the laser power is not varying during this measurement (as previously described in the power stability section). Measuring the intensity derived from an image is effected by the same factors that effect laser power.

Spectral Accuracy

Confocal spectral imaging (CSI) microscope systems are now in the market that delineates multiple fluorescent proteins, labels, or dyes within biological specimens by performing spectral characterizations. However, some CSI present inconsistent spectral profiles of reference spectra within a particular system as well as between related and unrelated instruments. There is also evidence of instability that if not diagnosed could lead to inconsistent and bad data. This variability confirms the need for diagnostic tools to provide a standardized, objective means of characterizing instability, evidence of misalignment, as well as performing calibration and validation functions. Our protocol uses an inexpensive MIDL that contains Hg+, Ar+, and inorganic fluorophores that emit distinct, stable spectral features, in place of a sample. By using an MIDL characterization, the accuracy and consistency of a CSI system is verified which will validate the acquisition of biological samples (8, 9).

We examined over 10 CSI systems (7 Zeiss Meta 510, 3 Leica SP1, 1 Leica AOBS, 2 Olympus FV1000) (Fig. 6). The Nikon C1si was not available to evaluate in the fall of 2005 but data provided by Nikon and other scientists that used the MIDL demonstrated that these system had a correct reference spectra for a properly functional spectral system. Most of these systems that were tested in the field with the MIDL lamp displayed spectral inconsistencies that enabled us to identify malfunctioning subsystems. By using a primary light source (MIDL) that emits an absolute standard “reference spectrum,” it was easy to diagnose instrument errors and measure accuracy and reproducibility of the system. Using this information, a CSI operator can determine whether a CSI system is working optimally and capable of making objective performance comparisons with the other CSI systems. If a CSI system were standardized to produce the same spectral profile as an MIDL lamp, researchers could be confident that the same experimental findings would be obtained on any CSI system. MIDL test can also be used to detect the misalignment of the system, instability, and possible existence of stray light.

thumbnail image

Figure 6. An Olympus FV 1000 confocal microscope was used to scan an MIDL lamp spectra using a separation of 2, 5, and 10 nm. An equivalent pattern was made from the PARISS system using 1 nm resolution. The features of the pattern, i.e., peak shape, peak, CV, minor peaks, and valleys, suggest the resolution of the system and the possibility of misaligned and unwanted reflections.

Download figure to PowerPoint

Spectral Test

  • 1
    The MIDL lamp is positioned on the stage. A low power dry lens is focused on the light bulb. If necessary for proper positioning, the microscope slide holder is removed and the light source position on the stage in close proximity to the objective.
  • 2
    A spectral scan consisting of 50 increments of 5 nm each was made between 400 and 650 nm using a 10× (0.4NA) Plan Apo lens without averaging and using an airy disk of 1 (Pinhole size 79 μm on a Leica SP1). It was also acceptable to use a 20× dry lens for this test. The efficiency of the light collection system was low necessitating that a high PMT voltage be used. It is not recommended to open the pinhole size as it will change the spectral pattern of the MIDL lamp.
  • 3
    The intensity of each 5 nm scan was calculated as the mean from a large ROI and the data were displayed as an intensity graph between the 400 and 650 nm range (Fig. 6).
  • 4
    The MIDL spectrum is dependent on the aperture size (i.e., 2 nm, 2.5 nm, 5 nm, or 10 nm) The patterns shown for the Zeiss 510 are not equivalent to the Leica (5 nm) or Olympus (2 nm) as five of the Zeiss 510 meta systems tested did not contain UV lasers and the Meta spectral scan began at 462 nm to utilize the laser in the machine. The units that contained UV lasers had spectral scans that started below 400 nm. The Zeiss LSM 510 system has fixed bins of 10.7 nm each and the system can measure eight bins simultaneously covering a range as of 86 nm simultaneously at the highest resolution. To cover the range of 462–650 nm, it is necessary to scan twice, each covering 86 nm. If there is insufficient light in the detection system from the lamp, the pinhole must be opened to acquire more light, but this will change the observable spectral pattern. The Nikon can scans 32 simultaneous channels and at a minimum setting of 2.5 nm, the total spectral scan would be 80 nm of bandwidth.
  • 5
    The PARISS system had a resolution of 1 nm, which yielded spectra that was similar to that obtained with both the Olympus and Nikon systems. The Olympus FV 1000 (Fig. 7) had a minimum resolution of 2 nm. The Nikon C1si has a minimum resolution of 2.5 nm resolution (Nikon data not shown, personal communication, Jeff Larson, and Jeremy Lerner).
  • 6
    To detect spectral accuracy above 650 nm, the lambda scan was repeated for the three Leica PMTs between the 650 and 800 nm regions. It was also repeated on the other units. The Nikon C1si is rated only to 720 nm while the other manufacturers supposedly have the ability to detect accurate spectra to 800 nm. This 650–800 nm scan can be made quickly on a cool MILD lamp as the bands in this region will disappear in a few minutes. The following particles were used to test the far red range: MP 760 nm fluorescent beads and Focal Check beads (MP, F36913, far red emission beads 692 nm) (8, 9). The far red beads can be excited with either of the following lasers: 543 nm, 561 nm, 568 nm, 633 nm, or 647 nm.
thumbnail image

Figure 7. Spectral scanning. The performance of an Olympus FV 1000 (A), Zeiss LSM 510 (B), and Leica AOBS confocal microscope (C,D) was measured using spectral scan functions and the light source (LightForm, Hillsborough, NJ) that was positioned on the stage in place of the laser light. The minimal band pass obtainable was used for each microscope; 2 nm for the Olympus, 5 nm for the Leica, and 10.7 nm for the Zeiss. The smaller band pass is responsible in part for the increased resolution observed with the Olympus microscope. In the Leica AOBS system, the PMT performance was measured across the 400–650 nm spectral ranges using the Leica spectral scan feature at two different time periods. C denotes that the three PMTs are yielding different spectra while D is suggestive of a machine that is correctly operating. The curves in C represent a machine that is out of spectral alignment. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Spectral PMT Comparisons

The spectral PMT performance was measured on the Leica SP1, SP2 and AOBS, the Zeiss LSM 510 Meta, and the Olympus FV 1000 using the MIDL lamp. It should be noted that the minimal slit opening varies from 2 nm to 10.7 nm in the spectral systems from the four companies. The spectral scanning of the MIDL lamp was made with a separation of 2 nm, 5 nm, and 10 nm using an Olympus FV1000 (Fig. 7). An equivalent pattern was made from the Pariss system using 1 nm resolution. These widths show the type of resolution that is obtainable with these various systems.

The three PMTs in a Leica system can be easily compared using the MIDL lamp with its many spectral peaks. To a less of a degree PMTs can be compared using either halogen light and bandpass filters or laser light with Chroma fluorescent slides. There are two types of PMTs used in the Leica system: PMT 1 is considered low noise (R6357) and PMT 2 and 3 (R6358) have high efficiency and sensitivity in the far-red wavelength regions. In our Leica SP system, PMT 2 has direct path to the detector while PMT 1, 3 needs to be reflected off an additional mirror prior to entering the detector. PMT 2 was superior to PMT 1, 3 for spectral imaging experiments. The test apparently shows not only the accurate representation of specific spectral bands, but the CV offers data on the system's functionality (5–9). The MIDL spectra should contain proper peak position registration with narrow peaks (small CVs) and deep valleys between the peaks. The PMT that has the best MIDL pattern in a Leica should be used for spectral scanning (Fig. 6). An MIDL pattern that is different from the expected pattern may reveal stray light, or misalignment of the system.

Spectrophotometer sliders and calibration: Mirror test.

The functioning of the the spectrophotometer sliders may be checked by the following two procedures. Set the machine up in reflection mode similar to doing the axial Z resolution test. Positioning of the sliders correlating to the 5 nm over the individual laser lines indicates the approximate relative position of the laser line. By moving the 5 nm slider above or below the laser line, the laser light should eventually be attenuated. The laser light should be reflected from regions 2–3 nm below and above the desired laser line of choice. If laser excitation light is reflected too far above or below the selected laser line, it suggests that the sliders may be out of calibration.

Figure 8B compares two ways to determine the spectral accuracy of the Leica confocal microscope. In the first method, a mirror is placed on the stage and the intensity of the reflection in adjusted so the peaks are approximately the same height. A lambda scan is made between 470 and 670 nm which encompasses the three lines. In a functional system, the FWHM was 5, 8, and 13 nm for the 488, 568, and 647 lenses. Measurements above that value suggest a system that needs calibration. This test is essentially the same as manually moving the slider above and below the laser lines. If the refection lines are wider than expected, it can also indicate that there is stray light in the system or interference from the laser light.

thumbnail image

Figure 8. Spectral scanning. A lambda spectral scan was made using a front sided mirror. The FWHM of the 488, 568, and 647 is 5 nm, 8 nm, and 13 nm, respectively. Higher resolution is obtained at shorter wavelengths. An MIDL lamp pattern was obtained on a Leica SP1 system as described in Figure 5A. Both patterns are indicative of system functionality. We feel that the MIDL lamp gives yields spectra from a defined line and shows how it is represented by the machine. In contrast, the lambda scan with reflection shows the channels in which data can be obtained and which channels can be omitted. It is believed that when the system is unstable or out of alignment both patterns (LightForm and Mirror) will be different.

Download figure to PowerPoint

This MIDL lamp is a defined standard that contain gases that emit a specific wavelengths that must register correctly in any spectrophotometer imaging system or the system needs calibration. The Light Form lamp can also be put on the stage and the pattern will show how the sliders are working in the system. The mirror reflection test can be compared with the MIDL lamp test. A lambda scan can be made with the MIDL lamp which shows defined gas peaks at 436 nm, 485 nm, 542 nm, 586 nm, and 612 nm. The position of the peaks, the shape of the peaks, and the values between the peaks yield information about the system spectral accuracy and performance of the system (5). Observed patterns that have narrow peaks with small CVs and FWHM that are located at the correct wavelength positions indicate that the sliders are functioning properly and the system is calibrated correctly. Patterns with choppy irregular curves or cutoffs at the high or low ends of the spectra can indicate the following optical problems: a malfunctioning slider, misaligned PMT, stray light, or general alignment problems.

The MIDL lamp can be used to measure the functionality of the system (8, 9). This test could have been described in the paper section on calibrating and validating the equipment. Because of the instability component observed with this test, it was put in this review chapter. Misalignment and malfunctioning of a component is illustrated with both Leica and Zeiss equipment in Figure 9. The three PMTs in a Leica SP1 are supposedly interchangeable in spectral scanning and are shown to have different MIDL spectral patterns. In Figure 9, PMT 3 shows a repression of the lower end spectrum and a shift of approximately 20 nm to higher values in the spectrum scan. Similar data has been reported previously on a different SP1 machine (8, 9). In a Zeiss 510 Meta, there is one channel (461 nm) that is not acquiring data. There is an also misrepresentation of the intensity of the lamp between 400 and 500 nm that was observed on both of the two Zeiss 510 multiphoton systems tested. The data in this region are higher than expected from the spectrum of other confocal systems shown in Figures 6–8. These data indicate that the Zeiss Meta system has an alignment problem, but since it is only one time point it cannot be determined that the system is unstable or just has a problem with the Meta detector.

thumbnail image

Figure 9. The MIDL lamp was used to show accurate representation of wavelengths of light in the Leica SP and Zeiss 510 systems. The Leica system has three PMTs that show different patterns. Two of PMTs (PMT 1, 2) are identical and one PMT (PMT 3) is shifted about 20 nm to higher wavelengths. It also has a suppression of the lower wavelengths. Clearly, PMT 3 has a misalignment problem. A Zeiss 510 Meta shows inaccurate representation of the 400–500 nm range and a suppression of one of the channels in this region. This test with the MIDL lamp demonstrated that both system need to be serviced to repair these problems. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Spectral System Instability

In flow cytometry, it is easy to run beads before, during, and after the experiment to measure the Coefficient of Variation of the bead population that ensures the fluidics have not changed during the course of an experiment. Can a similar test be used to show spectral stability using confocal microscopy? Even if everything is calibrated and checked out in accordance with the tests described above, the machine may still become unstable during the course of the experiment. The users can only determine this instability, if they are aware of the quality machine parameters they are measuring, use defined tests, and know what they should expect from the data they are obtaining. It is necessary to have a quick and reliable test to achieve this result. Often, it is necessary to test the machine at different hours during the experiment to ensure that the laser light sources have not shifted and an electronic component has not become unstable.

To illustrate this point on a confocal microscope, two scans of an MIDL lamp taken 4 h apart on a cold (stable T = 0) and a warm (unstable, T = 4.5 h) system (Fig. 10). Previously, we reported that MIDL patterns using different PMTs contained in the same system were not identical (8–10), but at the time, it was unclear why this phenomenon was occurring. These instability data were translated into bad image quality as illustrated with Red 6U Focal Check beads as there are extra reflections occurring in the same PMT with the same defined detection range (Fig. 11). The corresponding MIDL features of PMT 1 and 2 were also not identical and they were similar to that shown in Figure 10 for a warm and cold system.

thumbnail image

Figure 10. Instability. The MIDL showed a normal pattern for PMT 2 when the confocal system was just turned on (cold) but after 4.5 h of continuous use the spectra showed wider peaks and shifts in the position of the peaks. The MIDL spectral curves in the warm and cold systems no longer line up correctly, indicating a different spectral system. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

thumbnail image

Figure 11. Focal check beads. Correlated to the MIDL data in Figure 10, images of Focal Check 6 μm beads were acquired using a 63× lens on a cold system (T = 0) and a warm system (T = 4.5 h). Using the same detection range of wavelengths, excessive reflections were observed in the warm system, which is indicative of an unstable instrument. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

One extreme example of misalignment between PMT 1 and PMT 2 is illustrated using Muntjac skin fibroblast cells (Fluo slide #6) stained with Alexa 488, Alexa 555, and TOPRO 3. Using the same wavelength bandpass regions, it was observed that excitation with a 488 nm or 568 nm lasers yielded a good image in PMT 1, but a bad image in PMT 2 (Fig. 12). The image in PMT 2 contained many reflections and was acquired at a PMT voltage setting that was about 300 units below that using PMT 1.

thumbnail image

Figure 12. Cellular instability: Fluo 6 slide. FluoCells® prepared slide #6 (Molecular Probes, F36925) was used. Muntjac skin fibroblast cells stained with Alex 488, Alexa 555, and TOPRO 3. (See the methods section.) Cells were excited with the 488 laser line and measured in either PMT 1 or PMT 2 using the same bandpass range of 500–600 nm. The cells were also excited using the 568 nm laser line and measured in PMT 1 or PMT 2 using an identical bandpass range between 580 nm and 680 nm. PMT 1 demonstrated a correct representation of the sample. Using PMT 1, the 488 laser line excitation showed tubules, while the 568 nm laser line excitation revealed mitochondria as point sources of light. Using PMT 2, there was a combination of tubules, nuclei, and interference fringes with both excitation wavelengths. The PMT 1 was set to 702 volts for 488 nm excitation and 753 volts for 568 nm excitation. PMT 2 to be set at a very low voltage of 423 for both excitation laser sources because of the extra reflections of laser light in the system. Correlated with the bad data in PMT 2 were a shift in the MIDL spectrum similar to that shown in Figures 10 and 13 and a shift in the lambda scan shown in Figure 13. We speculate that these additional structures in the cells are due to interference errors or scatter at the slit-slider. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

thumbnail image

Figure 13. Instability. A mirror reflection test shows that PMT 1 region of acceptable wavelengths is broader than PMT 2 or PMT 3. On the same system, an MIDL lamp tests shows that PMT 1 has broader peaks than PMT 2 or PMT 3. The broadening of the peaks is suggestive of a system that is delivering less resolution and may be misaligned containing extra reflections. Systems that show bad MIDL patterns (bottom) or wide mirror reflection regions (top) will usually yield bad image data as shown in Figure 12.

Download figure to PowerPoint

This bad cellular image data (Fig. 12) was correlated to the data obtained from MIDL lamp (Fig. 8A) and the mirror lambda scan (Figs. 8B and 13) tests. A bad MIDL pattern or a widening of excitation reflections were indicators of system problems. A mirror reflection test shows that PMT 1 has a region of wavelengths that is broader than PMT 2 or PMT 3 (Fig. 13). The reflection scan also shows a broadening of the reflected laser light (Fig. 13). Both of these patterns are suggestive of a system delivering less resolution. The widening of the laser reflection may be due to the extra reflections or misalignment of the system.

How can the same sample yield two different images in the same system by only using two different PMTs contained in the same system? Apparently, one image is from an aligned calibrated light path (i.e., PMT1) and one image is from a misaligned light path (i.e., PMT 2). Since both PMTs were checked prior to the experimentation, it is assumed that something changed during the acquisition of these data. It is postulated that a board controlling the sliders overheated yielding incorrect data, but this is not confirmed. System instability is very hard to address as it can not always be duplicated during a service visit.

SUMMARY

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. SUMMARY
  6. Acknowledgements
  7. LITERATURE CITED

The instability can be detected by deviations from the normal expected values using specific tests that were designed to measure a functional system. In this paper, a few tests were described that yield data that are suggestive of an instable system. Other tests that have been described in part A of this review may also indicate instability. However, it is not easy to locate and describe a transient instability in a confocal system with any test.

The laser power instability shown in Figures 1–3 may possibly be attributed to the following factors: power supply, AOTF thermal regulation, improper thermal heat dissipation, fiber optical polarizations incompatibility, and electronic component failure. At this time, one definitive source of laser power fluctuations has not been identified. A recent replacement of the AOTF has eliminated most of the power fluctuations shown in Figures 1 and 3.

The galvanometer is the key component in confocal microscopy. It scans across the slide measuring fluorescence pixel by pixel. These pixels must be square and the grid should be absolutely straight with each scan being equivalent to the scan before and after it. A lack of galvanometer stability will definitely result in poor image quality.

A reference lamp (MIDL) has argon krypton gases that have a defined position in the spectrum. Deviations from the expected pattern suggest that the system is misaligned or unstable. This instability or misalignment can be also measured by Chroma narrow band pass filters, bead spectra, Chroma red fluorescence, and spectral separation (8, 9). Factors that can possibly influence the system to cause this instability include the following: fiber polarization mismatches with laser polarization, a defective operating AOTF, improper heat dissipation, defective slider mechanism, and defective heat sensitive electronic boards. The advantage of specific tests having known values and characteristic shapes is that they allow for the determination of normalcy and deviations from the state indicating either alignment errors or instability.

Some of the tests that have been described are very troublesome as they indicate that the confocal microscope system may have been operating correctly when the test was initially run but became unstable after continuous operation of the machine. How many times a day will one have to test a system to measure stability and understand how the changing parameters may affect their data? This concept of periodically checking equipment is similar to checking a flow cytometer in the AM and assessing that it is properly functioning only to find that it may become blocked in the PM and not fucntioning correctly. How many and what type of imaging parameters does one need to monitor to ensure that the system is functioning properly during the course of the day? Hopefully, a procedure will be developed other than just using a standard histological slide to assure the investigator that the machines are not becoming unstable during measurements and are delivering suboptimal data.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. SUMMARY
  6. Acknowledgements
  7. LITERATURE CITED

The authors thank Jeremy Lerner for providing them the light source to properly evaluate spectra in a confocal microscope and for his stimulating discussions. Figure 2 has been previously published in Cytometry and the journal has allowed this figure to be reproduced in this review article. The authors also thank the various scientists who have graciously allowed their machines to be tested. Their names have been withheld at the request of some of the scientists.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. SUMMARY
  6. Acknowledgements
  7. LITERATURE CITED