Independent Resolution Test of

Transcription

1 Independent Resolution Test of as conducted and published by Dr. Adam Puche, PhD University of Maryland June 2005 as presented by (formerly Thales Optem Inc.)

2 Independent Resolution Testing of OptiGrid Device (June 2005, as conducted and published by Dr. Adam Puche, PhD, University of Maryland) Introduction Structured illumination microscopy is a combination of optical and computational methods that remove out of focus information from an image. This method allows for optical sectioning, improved axial resolution (i.e. z-axis resolution), and improved lateral resolution due to the removal of out of focus light when compared to traditional epifluorescent imaging. These properties are currently most commonly achieved using a single or multi-photon confocal microscope setup. The purpose of this test was to compare the resolution improvements achievable with the OptiGrid implementation of structured illumination with the resolution improvements gained with a typical modern confocal microscope. Materials and Test Conditions The tests were performed with an OptiGrid (serial #0108) and a Q-Imaging Retiga EXi CCD camera (serial # A-138). The confocal microscope was n Olympus FluoView 500 single photon unit (manufacture date January 2004) equipped with 4 lasers and 4 photomultiplier tubes. This instrument is representative of what all current confocal microscopes can achieve. The epifluorescent illumination source was a standard Olympus lamp housing with a 103/W2 Mercury globe installed. Test conditions were setup to be as similar to the typical imaging environment of a biological specimen. Thus, all beads used in testing were prepared with a standard aqueous mounting media. The presence of aqueous mounting media, which has a slightly different refractive index to the glass coverslip, will introduce a slight spherical aberration that varies with imaging depth into the specimen. These tests assess the instruments susceptibility to this aberration which is typically seen an asymmetry in point spread functions. The testing was performed by Dr. Adam C. Puche, Ph.D., an Associate Professor at the University of Maryland. Dr. Puche has over 15 years experience with microscopy and runs a Digital Imaging and Confocal Microscopy core lab at the University of Maryland. The OptiGrid and camera were provided on loan for testing by Qioptiq Imaging Solutions (formerly Thales Optem Inc.). The results were prepared solely on the basis of testing using this demo unit and performed by Dr. Puche. No instructions or data beyond the operation instructions in the owner s manual were given by Qioptiq Imaging Solutions. Dr. Puche has no stock in Qioptiq Imaging Solutions or the parent Qioptiq Group, nor gains any financial benefits from sales of their products. Performance Tests Test #1: Point source imaging. The point source resolution capabilities of the OptiGrid, traditional epifluorescent microscope and confocal microscope were compared. The OptiGrid was calibrated following the manufacturers instructions. The aperture and other parameters on the confocal were set according to the manufacturer s Figure 1. Axial point spread measurement. The photographs show the axial images (XZ plane) taken with traditional epifluorescence, OptiGrid structured illumination, or a typical modern confocal microscope. The graph shows the numerical measurements of the intensity of out-of-focus light as the mean + standard deviation of seven different specks measured in both XZ and YZ planes. Page 1 of 4 All rights reserved, Adam Puche 2005

3 recommendation for the objective used. Green florescent microspheres from the Molecular Probes PS-Speck TM Microscope Point Source Kit were photographed using an Olympus 100x NA 1.4 oil immersion objective. The microspheres had a size range of nm and were photographed in an optical stack where each image was separated by 100nm in the z-axis. The image capture size (number of pixels) in the confocal was set such that the pixels/nm of the confocal matched the pixel/nm image size of the Q-imaging camera. Saturation of images was avoided by setting the exposure times and PMT values such that the brightest point in the image was upper 1/3 of the dynamic range of the camera/pmt (both had 12 bit dynamic range) and no pixels permitted to reach saturation values. The same microspheres were imaged with each instrument. The resulting images were scaled such that the brightest point in each image stack for each speck was set to 100% and the darkest point away from the speck set to 0% (Figure 1). The intensity distribution was measured in the XZ and YZ planes from 7 different specks and plotted as intensity vs. axial position (z-axis). There were no significant differences between XZ and YZ measures. The intensity vs. axial position profile for traditional epifluorescence (Figure 1, green trace) shows the expected distribution of extensive out-offocus light above and below the position of the speck (which is located at the most intense point in the curve). Traditional epifluorescence has a visible asymmetry in the distribution when imaging below the speck that is introduced by the refractive index difference between the coverslip and the aqueous mounting media as the image plane is moved through the specimen (Figure 1). Surprisingly the OptiGrid intensity profile (Figure 1, red trace) closely matches that derived from the confocal microscope (Figure 1, blue trace). The most superficial portion of the curve is not statistically different from that portion of the confocal microscope derived curve. This indicates that in an ideal optical environment the OptiGrid and confocal point spread functions are effectively identical. In the lower portion of the curve (where the aqueous media introduced spherical aberration is a factor) is still very close to what the confocal can produce. When comparing the asymmetry of the curves the confocal has a ratio of the 1.1 (ideal symmetry in the curve would be a 1.0 ratio), the OptiGrid has a ratio of 1.2 and the epifluorescent setup a ratio of 2.6 (distributions significantly different at P<0.05 between confocal and OptiGrid and P< between confocal/optigrid and epifluorescent using a MANOVA statistics test). This suggests the confocal is slightly better at handling the spherical aberration in a typical aqueous environment but both instruments deal with the effect significantly better than traditional epifluorescence. The overall OptiGrid performance on this axial imaging is only slightly less than that of the confocal microscope. Measured at the 50% intensity point of the distributions the width of the OptiGrid curve is 110% that of the confocal distribution with most of this difference due to the aberration introduced in the lower part of the curve. Both the confocal and OptiGrid show substantial improvement over traditional epifluorescence imaging (curve width at the 50% intensity point for traditional epifluorescence is 320% of the confocal). Lateral resolution (XY plane) of the OptiGrid is also similar to the confocal and significantly better than traditional epifluorescence (Figure 2(). Closely apposed individual specks can be resolved in both the confocal image and OptiGrid image when difficult to resolve in the epifluorescence image (Figure 2 insert panels). Figure 2. Comparison photographs of extended focus images in a region of densely packed specks. The OptiGrid and confocal microscope are both capable of resolving closely packed specks (insert panels). As expected there is significant haze from out-of-focus light in the traditional epifluorescent image.. Page 2 of 4 All rights reserved, Adam Puche 2005

4 Figure 4. A single optical slice through a 1µm FocalCheck TM bead that contains surface florescence (arrow) and a second bead that is uniformly fluorescent. The OptiGrid and Confocal both resolve the ring staining on the left bead. Figure 3. A single 6µm FocalCheck TM microsphere optically sectioned with each instrument. Shown is a view though the center of the sphere in the XY plane, and in the XZ plane through the image stack. TEST CONCLUSION: The OptiGrid implementation of structured illumination performs close to the capabilities of a typical modern confocal microscope. The measurable slight differences in performance when encountering the aberrations introduced by aqueous mounting media are unlikely to be noticeable when imaging biological specimens. Test #2: Optical sectioning. One of the most common uses of confocal microscopy is to optically section a tissue or cell to visualize internal details. To simulate and measure the performance of traditional epifluorescence, OptiGrid, and confocal microscopy FocalCheck TM fluorescent microspheres from Molecular Probes were imaged. These microspheres contain a thin ring of fluorescence at the surface of the sphere, thus a perfect optical section should resolve through the center of the spheres as a ring of fluorescence surrounding a dark center. Microspheres of 6µm (simulating the diameter of a small cell) and of 1µm (simulating a smaller structure such as a spine head) were optically sectioned on the OptiGrid and confocal microscope. As noted above both the OptiGrid and confocal were calibrated following the manufacturers recommended optimal settings and a 100x N.A. 1.4 oil immersion objective was used for all imaging. The confocal and OptiGrid both resolved the thin ring fluorescence surrounding the 6µm microspheres in optical sections through the microspheres (Figure 3). The slight asymmetry in viewing deeper into the specimens (described above) in the OptiGrid distorted the shape of the microsphere in the axial plane by ~5% compared to the confocal which had only ~2% shape asymmetry distortion. As expected traditional epifluorescence was incapable of resolving a spherical structure in the XZ plane, instead displaying an diffuse ellipsoid or teardrop shape due to the out-of-focus light above and below the microsphere (Figure 3). Using 1µm FocalCheck TM microspheres the confocal and OptiGrid were capable of resolving the ring fluorescence present on the surface of even this small a structure (Figure 4). The confocal images show slightly more grain (i.e. variations in intensity pixel to pixel) due to the nature of a PMT compared to a cooled CCD detector. This difference can easily be corrected with image averaging during acquisition, a setting commonly used with confocal microscopy, but one not used for these tests. Both instruments are capable of producing smooth low noise imagery. TEST CONCLUSION: The OptiGrid implementation of structured illumination performs close to the capabilities of a typical modern confocal microscope on optical sectioning. The measurable slight distortions in axial shape with the OptiGrid are unlikely to be a noticeable factor when imaging biological specimens. Page 3 of 4 All rights reserved, Adam Puche 2005

5 Summary The OptiGrid implantation of structured illumination microscopy can perform with optical sectioning characteristics extremely close and in ideal conditions identical to that of a modern confocal microscope. High resolution measurements show the confocal to have slightly superior optical sectioning capability when encountering refractive index changes in aqueous mounting media, although this difference is unlikely to be discernable when imaging biological specimens. The OptiGrid is a viable alternative to a confocal microscope for imaging discrete objects such as cells in biological specimens. Materials and Instruments details Molecualr Probes PS-Speck TM Microscope Point Source Kit (#P7220) Molecular Probes FocalCheck TM Thin Ring Microspheres (#F14791 and # F14808) Qioptiq Imaging Solutions OptiGrid (Serial #0108) Q-Imaging Retiga EXi CCD camera (serial # A-138) Olympus FV500 Confocal Microscope BX61 frame with confocal attachment Mercury lamp 103/W2 globe Olympus 100x N.A. 1.4 Objective (Model UPLSAP0100X0) MediaCybernetics Image Pro Plus (Version 5) Corel Corporation CoreDraw Graphics Suite (Version 12) Reindeer Graphics Image Processing Toolkit (Version 4) NCSS Statistics Package (Version 2003) Page 4 of 4 All rights reserved, Adam Puche 2005