Confocal Laser Microscope Scanner and CCD Camera

KAWAMURA Shin-ichiro¹ NEGISHI Hideomi¹ OTSUKI Shinya² TOMOSADA Nobuhiro¹

This paper presents the elemental scanner technology and CCD camera's characteristics that maximize the CSU10 Nipkow disk type confocal scanner unit. The scanner's internal structure that embodies the theory of forming confocal images at video rate is depicted; the system configuration, spatial resolution, and noise level of the CCD camera used with this scanner summarized; and examples of applications shown.

1. Motion & Measurement Business Div.
2. Sourcing & Manufacturing Business Div.

Introduction

Confocal scanning microscopes employing a Nipkow disk, an invention of Paul Gottlieb Nipkow of Germany in 1884, have delivered the merit of high-speed scanning, but the low light transmittance of pinholes of the Nipkow disk has proved to be a significant drawback. Use of a condensing lens is widely known as a classical and typical optical method for allowing a beam of light to effectively pass through a minute hole. Adopting a lens- array disk to apply this method to all the pin holes positioned in a spiral around a Nipkow disk, Yokogawa CSU10 confocal laser microscope scanner unit was introduced commercially for the fields of biological and biotechnological research and development.

Since its debut in 1996, nearly two hundred CSU10s have been shipped and many are serving in biotechnological and medical research laboratories. The CSU10 offers direct monitoring of an object with the naked eye, but its principle requires a camera to shoot the images and enable the use of those images as electronic data. To maximize the benefits of a system comprising the CSU10 and a camera, the focus is on examining the CSU10's characteristics and precisely matching them to the image-capturing system that includes the CCD camera.

Structure and Operating Principle of Scanner

Figure 1 Internal Structure of CSU10

In contrast to most conventional confocal microscopes which adopt single-beam scanning with oscillating light reflectors, the CSU10 employing a Nipkow disk adopts multi-beam scanning using about a thousand beams that are simultaneously emitted through a pin-holed disk to facilitate high-speed scanning.

Figure 1 depicts the internal structure of the CSU10. The laser beam is introduced by a fiber-optic cable engaging the scanner with an FC connector. After the beam is emitted from the fiber with divergence, it passes a neutral density filter (omitted from the figure). Reflecting the beam off the first mirror changes its course and it becomes parallel rays upon passing through a collimating lens. The beam passes through the slide-out excitation-wavelength select filter, and is then reflected by the second and third mirrors before passing through the micro-lens- array disk (hereinafter referred to as MLD), dichroic mirror (DM), pin-hole-array disk (PHD), and field lens (FL) in order. Since a general microscope is not telecentric in the image side and the principal beams of MLD have different angles, after the principal beams go through the microscope tube lens, the confocal images in the CSU10 would be bright at the center and dim away from the center without an appropriate choice of the FL. To solve this issue, an FL with a focal length is installed in the CSU10, so that the entire laser beam can pass the pupil of the objective lens.

Figure 2 Scanning in CSU10

The CSU10 is connected to the camera port of a microscope, and the laser rays is emitted from the CSU10 and led onto the specimen through the microscope objective. The signal light from the specimen runs the reverse optical path and reaches the DM of the CSU10. Since the CSU10 can be used for fluorescence observations as standard, the DM is designed so that it reflects fluorescence signals having a longer wavelength than that of the exciting laser. Consequently, the reflected rays are again reflected by the fourth and fifth mirrors, then led to the observation system. To prevent the excitation light from finding its way into the observation system, an interchangeable barrier filter is installed behind the DM. Accordingly, the confocal image can be observed through either the eyepiece or camera port by switching the slide mirror position.

Figure 2 illustrates the method of scanning in the CSU10. The MLD and PHD, precisely patterned in 250- μ m intervals, are horizontally opposed, mechanically coupled, and rotate at 1800 rotations per minute. The hole patterns are designed so that an image can finish being scanned by 30 degrees of rotation, making the image generation rate 360 frames per second. The incident rays (laser beam) are condensed by the individual micro lenses on the MLD, pass through the DM, then come into focus at the corresponding, opposed pinholes. Hence, the optical loss is minimized, the reflection at the PHD surface reduced, and thus signal-to-noise ratio increased. The pinhole plane is made optically conjugate with the object focal plane by the optical system of the microscope, so the many condensed laser beams scan the specimen on the object focal plane synchronously with the many pin holes being rotated and scanned. The fluorescence emitted from the specimen runs along the optical path of the microscope in the reverse direction and reaches the PHD. While a conventional confocal microscope requires a pinhole on both of the excitation and observation ends for each beam, the CSU10 uses the same pinhole, which results in excellent durability to mechanical vibrations. After passing through the pin holes, the fluorescence beams are reflected by the DM and transmitted to the observation system.

System and Camera

A computer is vital for most conventional confocal laser scanning microscopes since it is required to process the scanning position data and the intensity of the fluorescence signal, scanned by a laser beam with an oscillating mirror and acoustooptic device(s) and detected with a photomultiplier, into an image.

Figure 3 Scanning in CSU10

Figure 3 shows an example of systems using the CSU10. Even with the minimum system using laser scanning, the CSU10 allows confocal images to be observed through the eyepiece by the naked eye because it employs high-speed scanning and because of the persistence of human's vision. Nonetheless, since most cases require the observed images to be recorded, a CCD camera is used to record confocal images generated optically by the CSU10 as shown in Figure 3. This means that the recording performance of the CCD camera used determines the quality of the confocal images.

If the images captured by a CCD camera simply need to be recorded on videotape, a computer for image processing is not required. In most applications, however, an image capture board is installed in a computer, and the computer reads the images through the board, performs noise rejection processing and other image processing such as image enhancement and pseudo- colorization. Besides, most digital CCD cameras require a computer with an image capture device, and for Z-axis control for moving along the observation plane and automatic integrated control of the laser beam shielding shutter, even a Nipkow disk type scanner, such as the CSU10, requires a computer system with (1) a camera, (2) Z-axis control, (3) shutter control, and (4) image processing configured. Although items (2) to (4) are techniques that are also used for conventional laser scanning confocal microscopes, care must be taken when selecting item (1), the camera, since a two-dimensional sensor, which is not always required for conventional laser scanning confocal microscopes, is a requisite for capturing images with the CSU10. Furthermore, the quality of base images is the key determinant of success or failure in computer image processing, therefore the camera must be extremely carefully chosen.

There are numerous factors that discern the performance of a camera including the shooting speed, spatial resolution, sensitivity, noise level, number of gray scale levels, dynamic range, gamma, and color or black and white shooting capability. The users may select cameras having different speeds according to the purpose for use with the CSU10, and the spatial resolution, sensitivity, and noise level are often raised as the three common critical factors in actual applications. The spatial resolution is determined by the pixel size of the CCD camera; however, as in the case of noise level, data is rarely obtained under the same measurement conditions and hence the data of different cameras cannot be compared. To solve this, we devised a method of obtaining images shot by cameras under fixed conditions in which thin chromium films formed on a glass substrate in a pattern of repeated lines (four micrometers thick and four micrometers spaces) are shot while being lit from the back.

Performance Evaluation Measurements for Camera

Figure 4 Example of Performance Evaluation Measurements for Camera

Figure 4, an example of performance evaluation measurements for a camera, shows the profile of the gray scale in an image captured when shooting the chromium line pattern. An intensity level analysis was performed on the contrast and noise level of the image. The contrast is calculated according to the definition (I_W – I_B)/(I_W + I_B) that is generally used in optics where the mean intensities of the white and black levels are I_W and I_B, respectively, and the standard deviation of the gray scale distribution is calculated for each of the white and black level portions. The following describes the points for assessing each factor:

Spatial resolution: Although the pixel size, which is often shown in the catalog of a CCD camera, indicates the spatial resolution of the captured images for a digital camera, if the output from the camera is an analog signal, the spatial resolution also depends on the frame size of the image capture board. To use the CSU10 to perform video-rate or faster capture, a CCD camera with an intensifier may need to be used because of the low amount of light per unit time, which imposes a trade-off between sensitivity and resolution.

Sensitivity: Choosing a high-sensitivity camera, with the expectation of obtaining bright confocal images with the CSU10, results in images that on the whole, including the background, have high brightness. As the greatest benefit of confocal images is an increased contrast, the change in output electric signal level according to the unit change in the amount of incident light, namely the contrast, is also important for the camera. Especially for living samples, the dosage of a fluorescent reagent is limited and the energy of the excitation laser must be kept low to suppress the phototoxic effect from the laser radiation, therefore, the fluorescence signal is generally low in level and a camera with high sensitivity and high contrast is required. On the other hand, these limitations are reduced for non-living fixed samples but the same requirements also apply to camera selection when no treatment is performed to prevent discoloration.

The contrast obtained from the measurements in Figure 4 is 0.063, which is equivalent to 16-level gray scale for an 8-bit camera and 258-level gray scale for a 12-bit camera. The appropriateness of this value differs case by case; however, in general, moving-subject observations require a fast capture speed and hence the gray scale levels are traded off for higher sensitivity, but for morphologic observations halftones are required.

Noise level: Noise interposing in the obtained images incurs due to various optical and electrical causes, and its level differs depending on the camera's working conditions including the cooling temperature, set sensitivity, exposure time, capture speed, and extraneous noise. However, if these conditions are made constant, calculating the standard deviation of the gray scale distribution in the signals for each of the white and black level portions inside each captured image will enable quantitative comparisons and analyses. In the example of Figure 4, the noise level (standard deviation) in the white level portions is 6.15 and that in the black level portions is 2.63. Generally speaking, the noise level is less than 1 for cooled CCD cameras, 1 to 5 for non- cooled CCD cameras, and 5 or larger for CCD cameras with an intensifier. The criterion for whether the measured noise level is acceptable should be determined for each application. For example, it may be whether the finestructure of the sample's intended portion (e.g., the axon of a neuron) can be distinguished, or whether the changes in the sample's generating signal (sparks or fluorescence from pH-dependent granular pigments) can be identified and/or quantitatively measured. There is no camera available that boasts high spatial resolution, high sensitivity, and low noise, hence it is substantial to choose the right camera for the purpose in confocal image observations.

Examples of Measurements by CSU10

Figure 5 Confocal Images of Nematode through Continuous Z-axis Moves (taken by Professor Ayako Sugimoto, Biochemistry Dept., Science Div., Tokyo University, Japan)	Figure 6 Reconstructured Three-dimensional Stereoscopic Images of Nematode

 

Figure 7 Calcium Reaction of Hippocampal Nerve Fiber

Figure 5 shows images continuously shot while the microscope objective was moved in one-micrometer pitches. The sample is Caenorhabditis elegans (a nematode) in which a green fluorescent protein-fused pharyngeal protein is manifested. Since the CSU10 allows video-rate monitoring, three-dimensional slice data of a slow-moving nematode can be obtained with nearly negligible, spatial displacements. The results from obtaining slice image stacks at one-second intervals for sixteen seconds (namely, obtaining sixteen stacks) and performing three dimentional image restructuring processing for each stack using a computer, are shown in Figure 6. A CCD camera with an intensifier and an 8-bit image capture board was used.

Figure 7 shows video-rate images of a calcium reaction of a hippocampal nerve fiber at local dosing of glutamic acid. Althogh the figure shows confocal images shot at one-millisecond intervals, actual images were captured every one thirtieth of a second. The sample is a hippocampal nerve fiber stained with fluo-3, and photolysis of caged glutamic acid was carried out with an ultraviolet pulsed laser. As the camera, Model C2400-08 of Hamamatsu Photonics K.K. was used. It was observed that when a laser pulse was radiated at points A and B in turn at one-second intervals, a calcium wave propagated in both directions from the photolysis points (points A and B). In spite of the low-level noise interposing, this example shows that the camera used has sufficient resolution and sensitivity for observing how the calcium wave propagates.

Conclusion

A requirement for confocal scanners and image processing systems was fully automatic control including auto-matching with the digital camera's shooting speed, scanning speed control at speeds faster than video rate, and auto-filter switching. In 2001, we released a fully automatic scanner with a variable scanning rate from 1 to 33 milliseconds. We believe that this will expand the range of applications of confocal scanners in the realms of genetics, semiconductors, and liquid crystal displays.

References

1. Shinya Inoue and Kenneth R. Spring, Video Microscopy, 2nd Edition, 1997
2. Otsuki Shinya, "Realtime Image Observations with Confocal Laser-scanning Microscopy System," Igaku No Ayumi, vol. 186, no. 5, 1998, pp. 345-348, (Japanese)
3. Ichihara Akira and Kawamura Shin-ichiro. "Nipkow Disk Type Confocal Fluorescent Microscope," Hikari Alliance, vol. 8, issue 10, no. 97, 2000, pp. 53-59, (Japanese)
4. Kawamura Shin-ichiro, "Confocal Laser-scanning Fluorescent Microscope," Medical Photonics Course, Chapter 2, 2000, (Japanese)
5. Tomosada Nobuhiro, "Realtime 3D System," Seventh Summer Symposium of Confocal 488 , 1999, (Japanese)
6. Sakurai T., Terakawa S., Yamamoto S., Koshimoto H. and Watanabe A., "Glutamate-induced Degeneration of CA1 Neurons in the Rat Hippocampus Studied by Video Microscopy and Laser Photolysis of a Caged-compound" in "Slow Synaptic Responses and Modulation" (Kuba K. and Higashida H. ed.), 1999, pp. 361-366, Springer-Verlag Tokyo
7. Mitsuoka H., Unno N., Sakurai T., Kaneko H., Suzuki S., Konno H., Terakawa S., and Nakamura S. "Pathophysiological Role of Endothelins in Pulmonary Microcirculatory Disorders Due to Intestinal Ischemia and Reperfusion," J Surg Res. 87, 1999, pp. 143-152

• CSU10 is a registered trademark of Yokogawa Electric Corporation and other product and company names appearing in this paper are trademarks or registered trademarks of their respective holders.