This article has Open Peer Review reports available.
Extended Field Laser Confocal Microscopy (EFLCM): Combining automated Gigapixel image capture with in silicovirtual microscopy
© Flaberg et al; licensee BioMed Central Ltd. 2008
Received: 19 November 2007
Accepted: 16 July 2008
Published: 16 July 2008
Confocal laser scanning microscopy has revolutionized cell biology. However, the technique has major limitations in speed and sensitivity due to the fact that a single laser beam scans the sample, allowing only a few microseconds signal collection for each pixel. This limitation has been overcome by the introduction of parallel beam illumination techniques in combination with cold CCD camera based image capture.
Using the combination of microlens enhanced Nipkow spinning disc confocal illumination together with fully automated image capture and large scale in silico image processing we have developed a system allowing the acquisition, presentation and analysis of maximum resolution confocal panorama images of several Gigapixel size. We call the method Extended Field Laser Confocal Microscopy (EFLCM).
We show using the EFLCM technique that it is possible to create a continuous confocal multi-colour mosaic from thousands of individually captured images. EFLCM can digitize and analyze histological slides, sections of entire rodent organ and full size embryos. It can also record hundreds of thousands cultured cells at multiple wavelength in single event or time-lapse fashion on fixed slides, in live cell imaging chambers or microtiter plates.
The observer independent image capture of EFLCM allows quantitative measurements of fluorescence intensities and morphological parameters on a large number of cells. EFLCM therefore bridges the gap between the mainly illustrative fluorescence microscopy and purely quantitative flow cytometry. EFLCM can also be used as high content analysis (HCA) instrument for automated screening processes.
The introduction of laser scanning confocal fluorescence microscopy represented a major breakthrough in biology. The elimination of the disturbing out of focus light allowed the visualization of delicate sub-cellular structures that were hidden in the blur of thick samples [1–4]. Although the technique reveals structures with unprecedented sharpness, it also suffers from limitations. The images generate by scanning a sample with a single laser beam, where the in-focus signal passes a pinhole and is detected by a photomultiplier tube. This technique requires that the laser beam spends a few microseconds, depending on the signal to noise needed, on each and every point of the sample in order to produce a fluorescence signal. The size of the captured area is thus limited by the exposure time. To cover an area of 512 × 512 pixels the total exposure is in the range of 1000 ms, which is a relatively long time . Capturing an area over one Megapixel is rarely practical. To produce a sufficient amount of photons strong lasers are used that cause significant bleaching on fixed samples or potential harm to living cells.
An alternative to single-beam scanning systems is the use of a Nipkow spinning disc for out of focus blur elimination. The spinning disc has a spiral array of twenty thousand pinholes, producing a multi-beam illumination and thus allowing an instant confocal signal capture of an entire field. The efficiency of the illumination is significantly enhanced by the use of a multi-microlens disk in front of the pinhole array disk (the Yokogawa head design), which focuses the laser beams onto the pinholes . The major advantage of parallel beam illumination is that an image can be captured using cooled CCD cameras. Modern CCD cameras contain over a million individual photon-detectors with comparable quantum efficiency to photomultiplier tubes. The greatly increased speed and sensitivity of Nipkow spinning disc systems permits the use of low intensity excitation light resulting in reduced photo-damage [7, 8]. Due to this advantage, spinning disc confocal microscopy is currently a preferred choice for live cell imaging [6, 7, 9].
CCD cameras can rapidly generate immense amount of image information. This, combined with the explosive increase in the computational power of modern desktop computers, make large-scale image capture, processing and analysis available for routine research use.
Due to the inherent human bias in the selection of an image area, fluorescence microscopy has for a long time been considered as mainly an illustrative tool. However, there is a great demand for quantitative analysis of fluorescence properties in cells [10, 11] as illustrated by the extensive development of different flow cytometry techniques. Flow cytometry can objectively analyse hundreds of thousands of cells, but gives only limited and indirect (forward and side scatter) information about morphology. Already twenty-five years ago attempts were made to combine flow cytometry with a CCD-camera system producing still pictures showing preserved information of morphology and internal state of individual cells .
Automated, bias free image capture is the foundation of virtual microscopy. The general definition of virtual microscopy is the capture of a large amount of consecutive image tiles and the alignment of them into a continuous panorama, that is freely "zoomable" in silico [13–16]. Similar technology has for many years been used by astronomers to create panorama views of the night sky and by cartographers and military intelligence when examining series of satellite images of the earth. In pathology this is commonly achieved using line scanners with transmitted light illumination, where the linear-array detectors create a number of contiguous overlapping image stripes. During the last twenty years the development of several automated microscopy imaging techniques aimed the digitization of entire microscope slides and have led to the emergence of virtual histopathology .
An emerging field in biomedical imaging is the combination of fluorescence microscopy with automated capture techniques. Motorized XY-tables and high resolution cameras have been used to digitize entire slides using wide-field fluorescence illumination and 20X objectives . Fluorescence microscopy and automated mosaic imaging repeated over time has been used to study avian embryo development. The embryo was imaged using a 10X/NA 0,25 objective and the multi-field XY-area typically covered 8 images (875 μm × 688 μm), at 10 different Z-focus planes . In the field of automated identification and classification of cells it has been shown how to combine automated image capture with fluorescence microscopy for a high-throughput cell phenotype screening , . Here one of the goals and achieved result was to capture one cell per image in a mutlti-array/microtiter plate fashion without the need to reconstitute a mosaic of all wells .
A commercial example is the MetaMorph system, using wide-field epifluorescence illumination, executing automated multi-field image capture and the creation of large mosaics [21, 22]. Recently, original solutions with multi-objective arrays showed the possibility of fast wide-field capture over large areas . In 1999, early attempts at virtual confocal microscopy were made by manually stitching confocal laser scanning microscopy images in Adobe Photoshop . By the end of 2000, the power of automated Nipkow confocal microscopy and wide-field high-resolution cytometry was demonstrated in a new system. Due to the early version of the Nipkow module used, this system suffered drawbacks in illumination efficiency and was not sufficiently sensitive for low light fluorescence applications. Only 1–5% light throughput was reached, requiring long exposure times when using the confocal mode [25, 26]. In conclusion, there are several examples of how fluorescence microscopy is combined with automated imaging, but not to the extent of high resolution confocal microscopy, sensitive for low light fluorescence applications for areas reaching the Gigapixel size. We wanted to combine the microlens enhanced Yokogawa Nipkow spinning disc (40% light throughput ) confocal microscopy technique with automated image capture and analysis in order to produce high resolution images covering a very large area of interest, which could be analyzed automatically. Here we present the Extended Field Laser Confocal Microscopy (EFLCM) – method, where the end result is a Gigapixel mosaic showing confocal fluorescence information from several thousands of images at the maximal resolution achievable by a light microscope.
In this study we have used a custom built instrument based on the UltraVIEW LCI system (Perkin Elmer, USA). The system includes: A motorized Axiovert fluorescence microscope 200 M (Zeiss GmBH, Göttingen, Germany), a CSU10 Yokogawa head – a multilens/multi pinhole array aperture (Yokogawa, Japan), a motorized XY-table (Märzhauser, Germany), an ORCA ER cold CCD camera; detector array 1344 × 1024 px (Hammamatsu, Japan), a 3-line Argon-Krypton Laser; 488 nm, 568 nm, 647 nm (Melles Griot, USA) and a mercury lamp (Osram) for 365 nm illumination. The following objectives were used in this study (Zeiss):
10X/NA 0,25 Ph1 ∞/- A-Plan,
16X/NA 0,5 Imm Plan – Neofluar ∞/0,17
63X/NA 1,25 Oil Ph3 Plan – Neofluar ∞/0,17.
All hardware was controlled and automated using programs developed by us (see the Results section), in the programming environment OpenLab Automator (Improvision, Perkin Elmer). The raw multi-field image stacks were processed in and mosaics assembled using the ImageLab/Virtual Microscope software (Aragon System, Östersund, Sweden). The programs are available at the following web sites  and .
Preparation of biological samples
Both paraffin embedded and frozen slides of rodent tissues were labeled either by immunofluorescence or conventional histological (HE) staining. In vitro cultured cells were grown on cover glass in monolayers, alternatively suspension cells were centrifuged on glass slides using cytospin centrifuge (50.000/100 μl, centrifuged at 1500 rpm for 4 minutes). Immunofluorescence staining of viral antigens and transfection of monolayer cells with plasmids carrying GFP fusion proteins were carried out as described in earlier publication . The cells were fixed using methanol:aceton 1:1 mixture or 1% formaldehyde in PBS.
All cells used in this study were cultured in IMDM (Sigma) supplemented with 10% FCS (Sigma) and 50 μg/ml Gentamicin (Sigma). Cell suspensions were grown in a humidified incubator at 37°C in an atmosphere containing 5% CO2.
Automating the image capture process – the QuantCapture program
The multi-beam illumination of the microlens enhanced Nipkow spinning disc, the Megapixel image capture of cold CCD cameras combined with the possibility to automate image capture, inspired us to explore the possibility of generating confocal images over large areas. To achieve this we have assembled a fully motorized system based on an inverted fluorescence microscope equipped with a motorized XY-table. We have opted to control all motorized units of the microscope; illumination shutters, excitation/emission filter wheels and the different capture modes of the cold CCD camera. We control these units via QuantCapture4, a customized program that we have developed using OpenLab Automator. The OpenLab Automator is a high level, symbol based, visual programming language. It contains "action icons" for every aspect of hardware control, image handling, measurement functions, standard logical and arithmetic operations, as well as memory and file handling tools. These icons can be combined into complex networks, resembling flow charts, individually designed to execute intricate sequences of actions automatically. The Automator allows the declaration of numeric and string variables and variable arrays. It also allows the creation of subroutines that can exchange variables with the main program. The main program and the subroutines can then be linked into multi-task networks.
Moving Z capture
Confocal microscopy produces optical sections of focal planes that are free from out of focus blur. Inherently, however, it also restricts the collected information to the actual selected focal plane. To overcome this, microscopists regularly collect images from a series of individually captured optical sections at different Z-axis positions. These images are then examined individually or projected into a new summarized image using different projection algorithms (maximum intensity, sum, average, depth shading, surface shading etc ). High-resolution objectives (NA > 1) regularly mean high magnification and small field of view. Therefore, the capture of thousands of consecutive tiles is required to cover large areas at high resolution. Due to the fact that neither microscope slides nor mounted objects are perfect horizontal and no Z-motor work without a Z-drift, a capture over a large area often means that the sample is present in wide range of different focal planes. Generally different methods of auto-focusing are employed to keep the capture in focus over large areas. These algorithms are always relatively time consuming and also expose the sample for potential photo bleaching, without gaining information. To ensure that the sample stays in focus and to assure that all visual information of the sample is preserved, we have developed a capture method that we call the "Moving Z Capture", invoked in the second subroutine "Z Capture Loops" (Fig. 1C). Here we let the Z-motor move in pre-set steps from top to bottom focus along the Z-axis, while the cold CCD-camera collects all signals during one total exposure, without closing the shutter in between the Z-steps. Using this method, only one image consisting of all signals through the entire Z-axis is produced. The Nipkow spinning disc assures that only confocal signals from the different focal planes can reach the camera. When defining the capture area in the subroutine "Find Four Corners" (Fig. 1B), the program also prompts the user to define the top and bottom focus planes for the four corners of interest. The program calculates the absolute minimum and maximum Z-positions, shows the total focus range and prompts the user to decide how thick theoretical optical sections should be. The focus range divided by Z-step thickness gives the number of steps the Z-motor will move through and delay at with one given exposure time. The number of Z-steps will affect the total exposure time. The technique requires the exposition time of the camera to be precisely synchronized with the Z-motor travel time. We solved this challenge of synchronization by including a dummy capture cycle at every wavelength, which calculates the sum of the exposure times at every step together with the corresponding travel time of the Z-motor. This value sets the total exposure time for each wavelength in the "Z Capture Loops", the subroutine controlling the interactions of the Z-motor, emission and excitation filter wheels with the CCD camera. To correct for camera background noise, the program carries out automatic image subtractions using a "dark background" image. This "dark background" image is the noise the camera detects with closed illumination shutters, during one total exposure time.
In a Moving Z capture, the camera acts as an integrator, summing the image intensity at each pixel through the Z-depth travel. This produces a true sum projection of all signals present in the Z-depth/focus range. We found that the quality of a Moving Z image is better (higher S/N ratio) when compared to an ordinary sum-projection of individually captured Z-layers at identical Z-depth (see the figure MovingZcomparison, in Additional file 1). The exposure time at each step through the Z-stack is identical between the two types of capture techniques. The difference is that in the Moving Z capture the camera also collects signals even in between the steps, making the exposure time through the part of the Z-depth that actually displays the sample relatively longer compared to a single layer capture. Moreover, a sum-projection of individually captured Z-layers multiplies the background camera noise by the number of Z-layers. As we have measured, the background cold CCD camera noise contributes only once to the final image generated by the Moving Z capture. This background is also removed using the total exposure "dark background" image. The Moving Z technique has proved very appropriate for the purpose of the EFCLM method; to build mosaics in the XY-dimension with preserved Z-dimension information but without having to save individual Z-planes.
The program also allows periodic saving of the captured images, meaning that hard disc storage space is the only limitation of the captured area.
A modified version of QuantCapture4 has been developed to capture a full Z-depth image, at different excitation wavelengths from each well, in a 384-well plate.
The only requisite input value is the dimension of the captured area, as defined by rows and columns in the QuantCapture4 program. To address optical and mechanical imperfections inherent in an automated microscope system [15, 31], several image processing operations are used in the construction of the final, aligned mosaic. Potential imperfections corrected by the Virtual Microscope program are:
Imperfect XY-motor mechanics
Even a slight misalignment of the camera in relation to the position of the stage can generate significant difficulties when large arrays of images are captured. If the camera is not perfectly parallel to the stage a drift will accumulate and make it impossible to create an aligned mosaic. Rotating all the images prior to alignment can compensate for this defect. We found the angle of rotation by taking the median of all shifts necessary to align each image with its left neighbor. This means that in order to achieve perfect alignment in the stack it is necessary to first rotate and then register the images. The flat-fielding operation is also essential to enable alignment.
Visualization using in silico image browsing
We could demonstrate that the visualization features of theImageLab/Virtual Microscope software allow real-time browsing by building a presentation from the aligned stack "on the fly" (Fig. 2). Even with the high performance of modern desktop computers, a brute force implementation of stack visualization is not practical. To align and combine all image tiles into one huge, "browsable" mosaic would generate an image so big that an ordinary operating system would not be able to load it. The performance demands of the scaling operations involved in presenting the user view would also be prohibitive. As an example, a capture of 40 × 30 images with an individual size of 1200 × 1000 pixels would generate an image approximately 1.2 Gigapixels large (after cropping overlapping borders). If each pixel is captured in 3 color planes at a 12-bit resolution each – requiring 2 bytes of storage – the resulting image would be roughly 8 Gigabytes in size. To create a user-view we instead used a layered approach that both scales down the memory requirements of the OS and the performance demand on the application. The Virtual Microscope visualization application constructs the user-view by assembling image data from those tiles in the actual view. This mode is used whenever the magnification is high (Fig. 3B,F). At low magnification this strategy would be unnecessarily slow because a lot of tiles would have to be referenced to construct the view. Instead the view is then constructed from a moderately sized montage with maximum width/height of 5000 pixels. No observable details are lost using this approach because even the intermediate montage contains more details than a computer screen can present.
Image analysis and quantitation
Breaking the Gigapixel limit of a confluent confocal image
Examples of possible applications
The introduction of Nipkow spinning disc confocal technology, the ability to automate the actions of a motorized microscope and the increasing power of desk top computers have stretched the boundaries of microscopy and image acquisition. Combining these factors with advanced in silico image processing; we have shown that it is possible to create high-resolution confocal panoramas of specimens covering entire slides. Here we presented the Extended Field Laser Confocal Microscopy (EFLCM) method, allowing the presentation of a Gigapixel mosaic that combines information from several thousands of images showing both X,Y and Z-information.
In our view one of the most important emerging trends in microscopy is the automation of advanced microscopy techniques. Confocal microscopy is mainly used today as an illustrative tool. There is a need to use confocal microscopy as a technique for objective measurements. Automated, observer independent image capture can eliminate the greatest source of error in quantitative microscopy – the bias of the subjective field selection. It also produces images captured under identical conditions, preventing inhomogeneous exposure and differences in bleaching of the samples and thus allowing a more reliable quantitation of intensity analysis.
We have used the OpenLab Automator, a high level visual programming language, to create programs that drive automated microscope functions. Previously we have shown that hardware controlling programs created in a similar programming environment, ISee (ISee Imaging Systems, Raleigh), can be very useful in rapidly generating deconvolved 3D images . Through this publication we would like to advocate that high level visual programming languages enable users to tailor imaging applications, without extensive knowledge of complex computer languages.
To perfectly assemble very large arrays of images, with pixel precise matching, is however beyond the capacity of the OpenLab programming environment. To meet this challenge, we have developed the ImageLab/Virtual Microscope software. In this software the image stack is processed for imperfections and a new stack of perfectly aligned and intensity corrected images are constructed to create a seamless panorama. After this digitalization process the actual browsing of the specimen is independent of the physical microscope. Virtual microscopy has rapidly gained popularity in clinical pathology and medical teaching where bright field illumination allows the use of line scanners, permitting very rapid image acquisition in the Gigapixel range [16, 17, 41–45]. Although EFLCM uses a process of image capture (approximately 400 images/hour using a 63X objective, the time is dependent on the thickness of the z-depth, which increases with size of field) that is considerably slower than line scanners, its ability to detect very weak fluorescent signals, at different wavelengths and at the maximum resolution of a light microscope, gives it a great advantage. Moreover, the "Moving Z" confocal illumination technique generates images that quantitatively preserve all visual information from different focal planes thus allowing precise measurements within the context of preserved biological structures.
Adopting EFLCM to analyze 384-well plates can provide a simple instrumental solution for academic researchers helping to adopt high content analysis techniques in their everyday work. It may create a technological platform for large-scale analysis of living tumor or immune effector cells in routine clinical laboratory praxis to assist individualized assay guided therapy, as we have recently shown [37–39].
We have developed a technique that we call Extended Field Laser Confocal Microscopy (EFLCM). It is based on automated confocal microscopy and high performance image processing in silico. We have shown that it is possible to produce seamless confocal panoramas, using an image capture technique developed by us in the visual programming language Openlab Automator followed by advanced image processing and mosaic assembly in the ImageLab/Virtual Microscope software. We have shown that EFLCM is a useful tool for qualitative and quantitative analysis of histological slides, embryos, fetuses, a large number of transfected cells and of samples in 384-well plates.
This project was supported by grants from the Karolinska Institutet, Swedish Cancer Society (Cancerfonden), Swedish Research Council (Vetenskapsrådet) and from Cancer Research Institute (CRI), USA.
We would like to thank Doctor György Stuber for providing HHV8 GFP-LANA transient transfected cells and Tom Brass for his careful English language proof reading.
- Amos WB, White JG: How the confocal laser scanning microscope entered biological research. Biol Cell. 2003, 95 (6): 335-342. 10.1016/S0248-4900(03)00078-9.View ArticlePubMedGoogle Scholar
- Brakenhoff GJ, Voort van der HT, van Spronsen EA, Linnemans WA, Nanninga N: Three-dimensional chromatin distribution in neuroblastoma nuclei shown by confocal scanning laser microscopy. Nature. 1985, 317 (6039): 748-749. 10.1038/317748a0.View ArticlePubMedGoogle Scholar
- Zhang D, Wadsworth P, Hepler PK: Microtubule dynamics in living dividing plant cells: confocal imaging of microinjected fluorescent brain tubulin. Proc Natl Acad Sci USA. 1990, 87 (22): 8820-8824. 10.1073/pnas.87.22.8820.View ArticlePubMedPubMed CentralGoogle Scholar
- White JG, Amos WB, Fordham M: An evaluation of confocal versus conventional imaging of biological structures by fluorescence light microscopy. J Cell Biol. 1987, 105 (1): 41-48. 10.1083/jcb.105.1.41.View ArticlePubMedGoogle Scholar
- Lucitti JL, Dickinson ME: Moving toward the light: using new technology to answer old questions. Pediatr Res. 2006, 60 (1): 1-5. 10.1203/01.pdr.0000220318.49973.32.View ArticlePubMedPubMed CentralGoogle Scholar
- Saggau P: New methods and uses for fast optical scanning. Curr Opin Neurobiol. 2006, 16 (5): 543-550. 10.1016/j.conb.2006.08.011.View ArticlePubMedGoogle Scholar
- Wang E, Babbey CM, Dunn KW: Performance comparison between the high-speed Yokogawa spinning disc confocal system and single-point scanning confocal systems. J Microsc. 2005, 218 (Pt 2): 148-159. 10.1111/j.1365-2818.2005.01473.x.View ArticlePubMedGoogle Scholar
- Graf R, Rietdorf J, Zimmermann T: Live cell spinning disk microscopy. Adv Biochem Eng Biotechnol. 2005, 95: 57-75.PubMedGoogle Scholar
- Flaberg E, Stuber G, Szekely L: Multi-dimensional laser confocal microscopy on live cells in submicroliter volumes using glass capillaries. Acta Histochem Cytochem. 2006, 39 (4): 103-106. 10.1267/ahc.05056.View ArticlePubMedPubMed CentralGoogle Scholar
- Bocsi J, Varga VS, Molnar B, Sipos F, Tulassay Z, Tarnok A: Scanning fluorescent microscopy analysis is applicable for absolute and relative cell frequency determinations. Cytometry A. 2004, 61 (1): 1-8. 10.1002/cyto.a.20061.View ArticlePubMedGoogle Scholar
- Varga VS, Bocsi J, Sipos F, Csendes G, Tulassay Z, Molnar B: Scanning fluorescent microscopy is an alternative for quantitative fluorescent cell analysis. Cytometry A. 2004, 60 (1): 53-62. 10.1002/cyto.a.20027.View ArticlePubMedGoogle Scholar
- Yamamoto H, Tomioka , Atsuo : Flow cytometry and apparatus therefor. 1988, [http://www.freepatentsonline.com/4786165.html]Google Scholar
- Lundin M, Lundin J, Isola J: Virtual microscopy. J Clin Pathol. 2004, 57 (12): 1250-1251. 10.1136/jcp.2004.019919.View ArticlePubMedPubMed CentralGoogle Scholar
- Lee SH: Virtual microscopy: applications to hematology education and training. Hematology. 2005, 10 (Suppl 1): 151-153. 10.1080/10245330512331390249.View ArticlePubMedGoogle Scholar
- Glatz-Krieger K, Spornitz U, Spatz A, Mihatsch MJ, Glatz D: Factors to keep in mind when introducing virtual microscopy. Virchows Arch. 2006, 448 (3): 248-255. 10.1007/s00428-005-0112-2.View ArticlePubMedGoogle Scholar
- Leong FJ, McGee JO: Automated complete slide digitization: a medium for simultaneous viewing by multiple pathologists. J Pathol. 2001, 195 (4): 508-514. 10.1002/path.972.View ArticlePubMedGoogle Scholar
- Ying X, Monticello TM: Modern imaging technologies in toxicologic pathology: An overview. Toxicol Pathol. 2006, 34 (7): 815-826. 10.1080/01926230600918983.View ArticlePubMedGoogle Scholar
- Czirok A, Rupp PA, Rongish BJ, Little CD: Multi-field 3D scanning light microscopy of early embryogenesis. J Microsc. 2002, 206 (Pt 3): 209-217. 10.1046/j.1365-2818.2002.01032.x.View ArticlePubMedGoogle Scholar
- Mehes G, Lorch T, Ambros PF: Quantitative analysis of disseminated tumor cells in the bone marrow by automated fluorescence image analysis. Cytometry. 2000, 42 (6): 357-362. 10.1002/1097-0320(20001215)42:6<357::AID-CYTO1002>3.0.CO;2-6.View ArticlePubMedGoogle Scholar
- Conrad C, Erfle H, Warnat P, Daigle N, Lorch T, Ellenberg J, Pepperkok R, Eils R: Automatic identification of subcellular phenotypes on human cell arrays. Genome Res. 2004, 14 (6): 1130-1136. 10.1101/gr.2383804.View ArticlePubMedPubMed CentralGoogle Scholar
- Arrasate M, Finkbeiner S: Automated microscope system for determining factors that predict neuronal fate. Proc Natl Acad Sci USA. 2005, 102 (10): 3840-3845. 10.1073/pnas.0409777102.View ArticlePubMedPubMed CentralGoogle Scholar
- Narayan PJ, Gibbons HM, Mee EW, Faull RL, Dragunow M: High throughput quantification of cells with complex morphology in mixed cultures. J Neurosci Methods. 2007, 164 (2): 339-349. 10.1016/j.jneumeth.2007.04.016.View ArticlePubMedGoogle Scholar
- Weinstein RS, Descour MR, Liang C, Barker G, Scott KM, Richter L, Krupinski EA, Bhattacharyya AK, Davis JR, Graham AR, et al: An array microscope for ultrarapid virtual slide processing and telepathology. Design, fabrication, and validation study. Hum Pathol. 2004, 35 (11): 1303-1314. 10.1016/j.humpath.2004.09.002.View ArticlePubMedGoogle Scholar
- Beck JC, Murray JA, Willows AO, Cooper MS: Computer-assisted visualizations of neural networks: expanding the field of view using seamless confocal montaging. J Neurosci Methods. 2000, 98 (2): 155-163. 10.1016/S0165-0270(00)00200-4.View ArticlePubMedGoogle Scholar
- Kozubek M, Kozubek S, Lukasova E, Bartova E, Skalnikova M, Matula P, Matula P, Jirsova P, Cafourkova A, Koutna I: Combined confocal and wide-field high-resolution cytometry of fluorescent in situ hybridization-stained cells. Cytometry. 2001, 45 (1): 1-12. 10.1002/1097-0320(20010901)45:1<1::AID-CYTO1138>3.0.CO;2-M.View ArticlePubMedGoogle Scholar
- Wright SJ, Walker JS, Schatten H, Simerly C, McCarthy JJ, Schatten G: Confocal fluorescence microscopy with the tandem scanning light microscope. J Cell Sci. 1989, 94 (Pt 4): 617-624.PubMedGoogle Scholar
- Flaberg E, Szekely L: Openlab Automation – QuantCapture4. 2008, [http://188.8.131.52/EFLCM/Quantcapture]Google Scholar
- Sabelström P, Strandh C: ImageLab/Virtual Microscope software. 2008, [http://aragonsystem.com/]Google Scholar
- Stuber G, Mattsson K, Flaberg E, Kati E, Markasz L, Sheldon JA, Klein G, Schulz TF, Szekely L: HHV-8 encoded LANA-1 alters the higher organization of the cell nucleus. Mol Cancer. 2007, 6: 28-10.1186/1476-4598-6-28.View ArticlePubMedPubMed CentralGoogle Scholar
- Zamir EA, Czirok A, Rongish BJ, Little CD: A digital image-based method for computational tissue fate mapping during early avian morphogenesis. Ann Biomed Eng. 2005, 33 (6): 854-865. 10.1007/s10439-005-3037-7.View ArticlePubMedGoogle Scholar
- Sun C, Beare R, Hilsenstein V, Jackway P: Mosaicing of microscope images with global geometric and radiometric corrections. J Microsc. 2006, 224 (Pt 2): 158-165. 10.1111/j.1365-2818.2006.01687.x.View ArticlePubMedGoogle Scholar
- Kuglin CHD: The phase correlation image alignment method. Proceedings of the IEEE International Conference on Cybernetics and Society. 1975, 7: 163-165.Google Scholar
- Brown LMG: A survey of Image Registration Techniques. ACM Computing Surveys (CSUR). 1992, 24 (4): 325-376. 10.1145/146370.146374.View ArticleGoogle Scholar
- Brown LMG: Registration of multi-modal medical images: exploiting sensor relationships. Thesis (Ph D). 1996, Columbia University, 1996Google Scholar
- Leong FJ, Brady M, McGee JO: Correction of uneven illumination (vignetting) in digital microscopy images. J Clin Pathol. 2003, 56 (8): 619-621. 10.1136/jcp.56.8.619.View ArticlePubMedPubMed CentralGoogle Scholar
- Zhou X, Tian F, Sandzen J, Cao R, Flaberg E, Szekely L, Cao Y, Ohlsson C, Bergo MO, Boren J, et al: Filamin B deficiency in mice results in skeletal malformations and impaired microvascular development. Proc Natl Acad Sci USA. 2007, 104 (10): 3919-3924. 10.1073/pnas.0608360104.View ArticlePubMedPubMed CentralGoogle Scholar
- Markasz L, Kis LL, Stuber G, Flaberg E, Otvos R, Eksborg S, Skribek H, Olah E, Szekely L: Hodgkin-lymphoma-derived cells show high sensitivity to dactinomycin and paclitaxel. Leuk Lymphoma. 2007, 48 (9): 1835-1845. 10.1080/10428190701559132.View ArticlePubMedGoogle Scholar
- Markasz L, Stuber G, Flaberg E, Jernberg AG, Eksborg S, Olah E, Skribek H, Szekely L: Cytotoxic drug sensitivity of Epstein-Barr virus transformed lymphoblastoid B-cells. BMC Cancer. 2006, 6: 265-10.1186/1471-2407-6-265.View ArticlePubMedPubMed CentralGoogle Scholar
- Markasz L, Stuber G, Vanherberghen B, Flaberg E, Olah E, Carbone E, Eksborg S, Klein E, Skribek H, Szekely L: Effect of frequently used chemotherapeutic drugs on the cytotoxic activity of human natural killer cells. Mol Cancer Ther. 2007, 6 (2): 644-654. 10.1158/1535-7163.MCT-06-0358.View ArticlePubMedGoogle Scholar
- Holmvall P, Szekely L: Computer Programs That Allow Fast Acquisition, Visualization and Overlap Quantitation of Fluorescent 3D Microscopic Objects by Using Nearest-Neighbor Deconvolution Algorithm. Applied Immunohistochemistry & Molecular Morphology. 1999, 7 (3): 226-10.1097/00022744-199909000-00009.View ArticleGoogle Scholar
- Iregui M, Gomez F, Romero E: Strategies for efficient virtual microscopy in pathological samples using JPEG2000. Micron. 2007, 38 (7): 700-713. 10.1016/j.micron.2007.04.008.View ArticlePubMedGoogle Scholar
- Marchevsky AM, Khurana R, Thomas P, Scharre K, Farias P, Bose S: The use of virtual microscopy for proficiency testing in gynecologic cytopathology: a feasibility study using ScanScope. Arch Pathol Lab Med. 2006, 130 (3): 349-355.PubMedGoogle Scholar
- Dee FR: Virtual microscopy for comparative pathology. Toxicol Pathol. 2006, 34 (7): 966-967. 10.1080/01926230601123062.View ArticlePubMedGoogle Scholar
- Lundin M, Lundin J, Helin H, Isola J: A digital atlas of breast histopathology: an application of web based virtual microscopy. J Clin Pathol. 2004, 57 (12): 1288-1291. 10.1136/jcp.2004.018739.View ArticlePubMedPubMed CentralGoogle Scholar
- Demichelis F, Barbareschi M, Dalla Palma P, Forti S: The virtual case: a new method to completely digitize cytological and histological slides. Virchows Arch. 2002, 441 (2): 159-164. 10.1007/s00428-001-0561-1.View ArticlePubMedGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2342/8/13/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.