Content uploaded by Gábor Kiszler
Author content
All content in this area was uploaded by Gábor Kiszler
Content may be subject to copyright.
Digital microscopy – the upcoming revolution in histopathology teaching,
diagnostics, research and quality assurance
Tibor Krenacs
1
, Ivett Zsakovics
1
, Tamas Micsik
1
, Laszló Fonyad
1
, Sebestyen V. Varga
2
, Levente Ficsor
2
,
Gabor Kiszler
2
and Bela Molnar
2,3
1
Department of Pathology and Experimental Cancer Research
3
Department of Internal Medicine, Semmelweis University, Budapest
2
3DHISTECH Ltd, Budapest
Histopathologists play key roles both in diagnosing disease entities and determining biomarkers related to the prognosis
and response to specific therapy of malignant tumors. Histopathology is still firmly based on cell and tissue morphology
supplemented with in situ molecular information and these together can be studied through the optical microscope. Digital
microscopy creates the digital representation of the whole microscopic slides at decent quality, which can be dynamically
viewed, navigated and magnified through the computer monitor as driven with the mouse, and shared though computer
networks without spatial and temporal limitations. Digital slides can be integrated into existing hospital databases and
accessed through intranet or the Internet for teaching, primary diagnosis, teleconsultation and quality assurance. Discrete
pixels of calibrated qualities making up the slides allow automated image analysis and signal quantification for drawing
unbiased conclusions both in diagnostic and research applications. Also, the integration of digital tissue microarray (TMA)
slides and related sample data into common database permits high-throughput, validated studies of biomarker screening at
low cost and high standards. Therefore, utilization of the full power of computer technology for easily accessing multiple
functions and the Internet grants digital microscopy a great potential for upgrading the efficiency of the pathology
workflow and of pathologists. With resolving the critical issues, including standardization of data formats, secure and fast
internet communication and medico-legal aspects, digital microscopy is expected to play a revolutionary role in future
histopathology. This chapter aims to highlight the specific functions and benefits of digital microscopy in several fields of
this discipline.
Keywords: Digital microscopy; bright-field scanning; fluorescence scanning; histopathology teaching; telepathology;
automated image analysis; tissue microarray; high throughput cancer research
1. Introduction
Soon after the introduction of CCD (charge coupled device) cameras for image acquisition though the microscope in the
late 1990s, scientists started working on assembling series of digital microscopic images into giant montages for making
accessible whole microscopic slides through the computer monitor [1]. In the early 2000s digital microscopy, also
called virtual microscopy, suffered not only from inherent difficulties of pioneering a new technology but also from
lacking of proper information technology (IT) for handling of the hundreds of megabyte (MB) files representing a
microscopic section. By now, however, digital scanning technology, the software tools for digital slide management and
IT support including computer networks have matured into the stage that makes digital microscopy an attractive option
for daily use in morphology-based disciplines particularly in histopathology (surgical pathology) [2]. Considering the
extra features digital slide technology adds to those of traditional optical microscopy and the expected advancement in
computer technology including speed, storage capacity and network accessibility, one can clearly envisage a digital
revolution in histopathology within the next 5-10 years where digital microscopy will gain extensive use in every aspect
of this discipline.
By studying stained tissue sections under the microscope surgical pathologists can reveal structural and molecular
alterations in disease processes to determine diagnostic entities and their biological behavior. The integrative role of
histopathologists particularly in diagnosing malignant tumors and screening for biomarkers related to patients’ response
to molecular targeted therapy upgrades their responsibility in therapy decisions [3, 4]. Though the recent progress in
molecular technology has allowed better understanding of the biological background of disease processes, tissue and
cell morphology still plays major part in the diagnostic process [5]. Techniques of molecular morphology combining
structural and molecular information are of increasing importance including genomic in situ hybridization (ISH) for
detecting chromosome and gene abnormalities [6], and immunohistochemistry (IHC) for revealing translated proteins in
normal and diseased cells [7]. In several cases of tumor pathology, series of biomarkers need to be determined semi-
quantitatively at the cellular or even subcellular level within the tissue architecture [8]. The expected high diagnostic
standards and the amount and complexity of information to be handled require support from artificial intelligence of
computing power. Hospital administration and diagnostic reporting systems already exploit digital data management,
which integrate alpha-numerical data, still digital images of macroscopic samples and radiology images, and voice
recognition [9, 10]. Digital whole-slides can be perfectly integrated into such digital database systems, provided that
appropriate computing and network power are available for handling the enormous size of digital slides.
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
©FORMATEX 2010
965
___________________________________________________
Digital microscopy offers unique features which are not available for conventional optical microscopy [2, 11].
Assisted by dedicated software tools it permits dynamic and prompt access to any detail of stained slides at arbitrary
microscopic magnifications as controlled with a mouse through the computer monitor. Digital slides of tissue sections
or cells can be shared through computer networks by many pathologists worldwide without spatial and temporal
restrictions. This unlimited access of slides offers digital microscopy as an efficient tool for telepathology yielding
primary diagnosis, teleconsultation for second opinion, graduate teaching, continuous education, proficiency testing,
external quality assurance and interlaboratory process validation [11, 12]. Furthermore, the calibrated qualities of
discrete pixels forming a digital slide permit automated image analysis and quantification.
Our group at 3DHISTECH in cooperation with the 1
st
Department of Pathology and Experimental Cancer Research,
Semmelweis University, Budapest has developed software and hardware tools for the widespread utilization of digital
microscopy. These are used in this chapter along with other published examples to demonstrate the potential of this
technique in histopathology. It is important to emphasize that digital slides will never replace conventional glass slides.
Rather, they allow highly efficient access and exploitation of the inherent information from them. As a consequence,
digital microscopy can substantially enhance the efficiency and accuracy of histopathologists and the whole
histopathology workflow for the overall benefit of the healthcare service [13].
2. Digital whole slide scanning – theory and practice
Digital microscopy creates large digital files representing all crucial details of stained tissue sections with decent
resolution and high color fidelity achieved using automated focusing and white balance [1]. Digital slides are made up
of giant arrays of rectangular pixels organized along x-y coordinates, each of which is characterized by size, color and
intensity values. Produced either by area or line scanning (see below) digital slides are built up as pyramids of
microscopic image series where low power views are generated by compressing the original sharp and optimally lit
images (Fig. 1A). Scanning through several focus levels within the usual 3-8µm sample thickness offers access to the z-
dimension used for emulating fine focusing of the optical microscope [14].
Fig. 1 A) The image pyramid generated from optimized x20 magnification image series by compression allows viewing of the digital
slide at arbitrary magnifications (e.g. x15 and x10). Scanning at different focus levels within the sample thickness (~3-8µm) offers
access to fine details in the z-dimension. B) Relative comparison of conditions of using different objective lenses for scanning the
same slide area. Using x10 lens offers high field of view (FOV) size and focus depth, while requiring small storage space. x20
objective allows double of the optical resolution than that of x10, but on the expense of revealing smaller FOV and focus depth while
increasing the storage need. x40 objective does not offer significant improvement in optical resolution compared to x20 (NA=0.9 vs.
0.8) despite requiring large storage space and long scanning time.
The spatial image resolution is primarily determined by the optical resolution characterized by the numerical aperture
(NA) of the objective lens and the pixel density of the CCD camera or sensor, which need to be matched for optimum
performance. Digital microscopes mostly use high NA (0.8) x20 objectives except for scanning cytology samples where
x40 objective offers somewhat better resolution (NA=0.9), however, this is on the expense of multiplying image size
and scanning time four times and narrowing the focus depth (Fig. 1B). Considering that the physical pixel size of a high
standard (1360x1024 pixel) CCD camera is 4,65µm, a x20 objective with x1 C-mount adapter can create a digital
resolution of 0,23 µm (4,65/20x1), which is near the Abbe-limit of resolution at the horizontal axis when using visible
light (~400-750 nm), and cover a 1360x0,23=312,8 µm image width. Advanced digital microscopy allows scanning in
both bright-field mode for emulating conventional transmission microscopy, and fluorescence mode for substituting an
epifluorescence microscope supported with up to 9 emission filters.
Photons emitted by the sample are collected either by photodiode CCD or complementary metal oxide semiconductor
(CMOS) sensors and translated into digital signals. Available slide scanners use motorized stage with continuous
autofocusing and utilize either the area scanning or the line scanning principle [15] (Fig. 2).
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
966
©FORMATEX 2010
______________________________________________
Fig. 2 Schematic representation of area scanning (A,B) and line scanning (C,D) techniques used by available slide scanners. A)
Classical area scanners collect large series of images at x-y dimensions through a microscope objective with a CCD camera either in
bright-field or fluorescence mode. B) The area scanner combining an 80-element lenslet array with complementary metal oxide
semiconductor (CMOS) sensor can cover large section areas at once. C) Typical line scanners can collect image strips from the
continuously moving slides through an objective using a linear array light sensor, which is however, not sensitive enough for
fluorescence signals. D) Combination of 64 or more of linear array sensors permits TDI (time delay and integration) scanning, where
consecutive sensors cumulate the signals making TDI appropriate also for fluorescence scanning.
In area scanners such as the Pannoramic Scan (3DHISTECH Ltd. Budapest), hundreds or thousands of adjacent
microscopic field-of-views (FOV) are rapidly taken by a CCD camera and aligned with software using stitching
algorithms which make seamless borders between tiles. The “array microscope” of DMetrix Inc. (Tucson, Arizona)
works under the same principle except extending the size of FOV by using an “80-element lenslet array” instead of a
single objective [16]. Typical line scanners such as the ScanScope (Aperio Technologies, Inc., Vista, CA) use a 4096-
pixel linear array light sensor, which is triple of the horizontal pixel number used by a CCD camera, and continuously
scans strips of the sample at this width line-by-line and produce much less image edges to stitch than area scanning
[14]. However, the scanning speed of line scanners is restricted by their low photon collecting capability, which also
prevents them from detecting fluorescence signals that usually require longer exposure time. TDI (time delay and
integration) sensors such as the one used by the Nanozoomer (Hamamatsu Inc., Hamamatsu, Japan) synchronize 64 or
more of these linear array sensors into a 2-D matrix thus combining high speed and light sensitivity. Recently,
successful efforts has been made to upgrade area camera scanners from stop-and-go scanning mode into continuous
motion scanning, which makes them highly competitive even with the most advanced TDI scanners.
3. Unique features of digital slides compared to traditional microscopy
Viewing slides though the computer monitor with easy access to the multi-functionality of a computer is much more
ergonomic than peeping though the ocular of an optical microscope. Experience shows that even pathologists with high
affection to conventional microscopy respect the benefits of digital microscopy if they spend enough time on practicing.
The computer generated image pyramid format of digital slides allows in-focus navigation through continuously
changing magnifications without the need for changing objectives, realigning the focus or the lighting conditions [11].
Digital magnifications beyond that used for scanning could still reveal fine microscopic details hidden at the original
magnification. Slides can be tilted arbitrary for proper orientation and preview images of the whole slide are available
simultaneously on the monitor where navigation history of high power analysis can also be traced (Fig. 3).
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
©FORMATEX 2010
967
______________________________________________
Fig. 3 A) Digital slide viewer interfaces utilize the whole computer monitor where preview images and navigation history (left side)
of high power analysis can also be traced. Calibrated pixels allow straight measurements of object distance, perimeter or area
highlighted by permanent annotations. B) The monitor can be shared for several digital slides for comparative studies as shown by
the same area of serial slides stained for H&E (left), the proliferation marker Ki67 (brown; middle) and the gap junction connexin-43
(red immunofluorescence) combined with the Ki67 protein (green; right), respectively in oral epithelial hyperplasia.
Permanent annotations and text put on digital slides, straight measurements of object distance, perimeter or area and
prompt still-image archiving at publication quality all support the pathology workflow. Several digital slides can be
opened side-by-side on the monitor for comparative analysis of serial slides of a sample stained for different
biomarkers. Even samples of immunohistochemistry and FISH (fluorescence in situ hybridization) can be opened,
linked and navigated alongside, which normally require the use of consecutive steps or even separate microscopes.
The pixels making up the images have calibrated dimension, color and intensity which features can be used for color-
separation based automated quantification and measurements of image-objects [15]. Furthermore, pattern recognition of
morphological and functional units within the tissue, such as glands, or hyperplastic or abnormally arranged epithelial
nests can be automatically made based on shape, size and texture identification [17]. Serial digital slides can also be
assembled into a 3-dimensional structure for reconstructing tissue architecture, e.g. for studying tumor invasion or
reorienting colorectal biopsies [18]. Digital information, including whole digital slides with their annotations and
measurements can be integrated into digital databases and shared though intranet or the Internet with unlimited partners
even at the same time. The freedom in accessing digital slide archives for re-reviewing and the logistics of slide storage
and sorting have become simple tasks managed through the computer. All these, of course require, advanced
information technology including high speed computers, massive storage capacity, safe data handling using backup
storage, along with wide-range Internet access and dedicated software tools offering a user-friendly graphic interface.
4. Immunofluorescence and FISH applications and the detection of small signals
Fluorescence microscopy detects fluorophores used for labeling molecules in cells and tissues with the techniques of
molecular morphology (see above) [19]. Fluorophores are activated at UV or visible wavelength to emit light of lower
frequency, usually in blue, green or red, which can be collected through emission filters in dark background. Samples
targeted with fluorescing labels particularly genomic FISH must be studied within a short time-frame to avoid false
negative results due to rapid signal fading (fading artifacts). In addition, small signals of a few hundred nm size, such as
those of gene and chromosome probes in FISH or those of its chromogenic version (CISH) are randomly placed
throughout the whole 3-8 µm thickness of tissue sections or cells and thus some remain hidden from the conventional
single focus photography [20, 21]. These signals can only be revealed with confidence by scanning several focus planes
through the sample (z-stacking, or extended focus) for proper quantification of gene/chromosome gain such as that of
HER2 on chromosome17 (CEP17) in breast cancer (Fig. 4); or for the fine spatial localization of signals proving gene-
translocations, such as that of t (9;22) resulting in the BCR-ABL gene fusion in chronic myeloid leukemia (not shown).
Gene abnormalities may determine specific diagnosis and the concomitant treatment options in increasing number of
malignant tumors [22].
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
968
©FORMATEX 2010
______________________________________________
Fig. 4 Some FISH signals of HER-2 gene (red) and CEP17 (green) of less than a micron size remain hidden from single focus
photography (A), but can be revealed when multiple focus layers are scanned and then projected (B; extended focus). Please note that
several green signals that are missing from A are clearly seen on B. C) Accumulation of all red and green signals gained from
merging consecutive optical layers allows reliable analysis. D) FISH (fluorescence in situ hybridization) signals revealed in a cell can
be intensity amplified in 3D (E) for better assessment. F) Cell nuclei can be automatically sorted into groups in a gallery according to
their FISH pattern and re-localized into their tissue environment.
Of the scanner types only area/tile scanners or TDI scanners are appropriate for fluorescence scanning. Considering
the need for routinely scanning several focal planes at numerous color channels, whole-slide scanning of fluorescence
samples represents a major challenge in digital microscopy. All separate channels and layers to be scanned multiply the
scanning time and storage space required. Selection of the objective is another crucial consideration in these respects.
The higher the optical resolution of an objective the better it detects small and low intensity signals, however, resolution
is inversely proportional with the focal depth an objective can depict sharp at z-dimension (see Fig. 1B). Though x40
objective has somewhat better chance than x20 objective to resolve small signals, it requires four times of the scanning
duration and storage space, and more optical layers than that of x20 objective for scanning the same slide. Using high
NA (0.8) x20 objective offers an acceptable compromise for revealing the majority of FISH signals with scanning
relatively limited number of z-planes within reasonable scanning duration and storage requirements. Signal detection
sensitivity can be further improved with applying a wide dynamic range cooled monochrome CCD camera and light
emitting diode (LED)-based fluorescence excitation [23].
The constrains of conventional fluorescence microscopy including rapid fading and limited area of archives make
large scale fluorescence studies, which in addition involve the handling of large data sets, extremely complicated. In
addition to the listed benefits of digital microscopy, digital fluorescence scanning allows every focal layer and color
channel to be reviewed or even re-colored for better signal separation (contrast) or combined arbitrarily, similar to that
in laser scanning microscopy. Though the resolution of digital microscopy is behind of that of confocal laser scanning,
it offers large areas instead of FOVs and a continuous range of analysis between low and high power views, thus it
represents a unique option for large scale fluorescence studies. By combining fluorescence digital microcopy, TMA
technology and project-based digital data-management, validated high-throughput studies can be performed (see later)
even for small and/or low intensity objects such as those detected with FISH or those highlighting gap junction
connexin channels with immunofluorescence (see Fig. 2). The linearity of digital fluorescence signals supports the
consistency of automated image analysis (see later) and slide-based image cytometry [23]. Since image data and the
integrated measurements and annotations are permanent, validation of datasets by other pathologists for quality
assurance purposes is straightforward and easy.
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
©FORMATEX 2010
969
______________________________________________
5. Digital/virtual slides for pathology teaching and continuous education
The most prevalent application of digital microscopy is its use for teaching and continuous education of anatomical
pathology, which is justified both by its versatility and reasonable cost [24]. The number of local computers served and
their location may differ but basic system requirements for teaching is very similar to that used for online intranet
teleconsultation (see later). The cost saving is obvious when one considers the burden of purchasing and maintaining
optical microscopes and the regular replacement of tutorial glass slides. The most obvious benefit of digital microscopy
for teaching lies in the access by everyone to the best selection of standard slides without spatial or temporal limitations.
Furthermore, a single digital slide can demonstrate a rare entity or any expensive ancillary technique such as
immunohistochemistry or FISH to serve unlimited number of students. There are many additional rewards. For
instance, parallel viewing of two or more slides allows opening several copies even of the same slide and correlating
distant areas at high power simultaneously. Annotation labels and text help guide students at distant learning and self-
assessment or can be requested at written exams using digital slides when students’ skills are tested. Appropriate
software can help randomize test orders for students sitting side-by-side at an exam and allows automated assessment of
multi-choice tests. In addition, computer technology extends teaching into a multi-partner online consultation as
opposed to the traditional one-to-one interactions at the optical microscope.
Following a validation process, digital slides have been used by our group for histopathology teaching since 2007 at
the 1
st
Department of Pathology and Experimental Cancer Research, Semmelweis University, Budapest [25]. More than
1000 hours experience in digital teaching shows high acceptance of the technology by graduate students and residents,
in line with published data by other groups [26]. Our survey asking over 200 students revealed that image quality was
the best ranked feature, while slide accession speed was the most criticized factor in a 40-computer intranet set up,
when running at full capacity. For teaching cytology, hematology, microbiology and urine analysis, where intracellular
resolution can be critical, samples scanned through several focal planes are preferred [27]. In our teaching module over
200 digital slides of the complete histopathology curriculum are stored on the external www.pathonet.com server and
are freely accessible for students from home without hiring a microscope. In 2009, 97% of students accessed the remote
slide box by initiating ~100.000 page loads mainly around the exam period, which means a massive ~300 page
loads/student considering that we take 350 students every year. Similar free-access web sites for histopathology
teaching are run for instance, at the University of Achen (www.vm.rwth-aachen.de), the University of Basel
(www.pathorama.ch) and the University of Iowa (www.path.uiowa.edu/virtualslidebox).
Digital slides can also be embedded into full text educational material. An example of such E-Book focuses on
immunophenotypic markers of melanoma (http://www.pathonet.com/ index.php?module=node&code=melanoma_kt)
and serves for continuing postgraduate education. Digital slide collections of national and international slide seminars
for self assessment training and proficiency testing are also accessible through this site. The College of American
Pathologists (http://www.cap.org) also offers online digital slide programs for general surgical pathology and
subspecialties such as dermatopathology or fine needle aspiration cytology. A web-based digital atlas of breast
histopathology (http://www.webmicroscope.net/breastatlas) has been proven also as an excellent tool for both graduate
and postgraduate education [28].
6. Diagnostic telepathology, teleconsulation and quality assurance
The enormous amount of morphological, clinical and molecular knowledge accumulating about disease processes
recently has made histopathology a highly sub-specialized discipline. The number of centralized diagnostic departments
staffed to cover all the subspecialties including e.g. hemato-, neuro- or soft tissue pathology etc…, are rare. The vast
majority of histopathology laboratories are of small or medium sized, which have to rely on external consultations with
specialist pathologists [2]. In addition, several laboratories, particularly in remote areas, or in developing countries are
understaffed and/or under-equipped and need regular outside help to meet all their primary diagnostic requests.
Traditional consultation involves posting of glass slides to external expert, sometimes consecutively, with the inherent
risk of loosing or damaging slides through transportation and gambling with patients’ chances for getting the right
treatment in time. Telepathology aims to support prompt consultations by communicating all data digital.
Static telepathology using still images suffers from sampling bias as opposed the free navigation on the entire digital
microscopic slide. Ideally, dynamic telepathology systems allow remote access by pathologists to digital slides and
relevant clinical/radiological data either online or offline, and offer a user-friendly computer interface for analysis and
reporting, which then can be integrated into hospital databases. An important attribute of dynamic telepathology is that
it offers equal opportunities for professional of restricted mobility including disabled people or those who are on
childcare. In such dynamic systems high throughput auto-feeding slide scanners continuously transfer digital slides into
servers to be accessed from computers either though intra- or the Internet (Fig. 5). The feasibility of using digital
microscopy for diagnostic purposes is a highly complex issue determined by several technical, professional and medico-
legal aspects. These are related to scanning/server and network speed, image quality, storage space, appropriate backup
system, the quality of workstation and graphic interface, technically trained professionals, internal method validation,
ease of communication with the hospital database and secure data transfer protecting patients’ rights [29, 30, 31, 32, 2].
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
970
©FORMATEX 2010
______________________________________________
Current diagnostic applications of digital microscopy include intra-operative frozen section services [33], routine
histopathology services [34, 35] and teleconsultation either for a second opinion or subspecialty consultations [36].
Fig. 5 Basic setup of a telepathology system. The high throughput slide scanner transfers digital slides into the internal image server
(1) to be accessed along with the data server from departmental computers (pathologist 1,-2, etc…) though the intranet either for
telepathology, -consultation (2) or teaching (3). A digital pathology teaching setup using dedicated software with 40 computers
linked through the intranet is run at the 1
st
Department of Pathology and Experimental Cancer Research Semmelweis University,
Budapest (right). Digital slides are also continuously uploaded from the internal to an external image server through the institutional
Firewall (4), to be accessed through the Internet by expert pathologists for telepathology, -consultation from anywhere around the
globe.
At present, the main bottleneck of dynamic digital (also mentioned as virtual) telepathology involves the network
transfer of large image files, which is further complicated by the obstacle of security filtering of data by hospital
firewalls. Installing the image server external to the hospital firewall for passing through only one security filter at up-
or downloading, and working only with one user-selected fragment (area of current interest) of the slide at a time can
help moderate these difficulties. An optimized multi-server connection set up among five European academic centers
has been proven to double the viewing speed to achieve 3.1Mbit/s, when remotely accessing digital slides from other
centers [31]. Settling the issue of adequate data propagation needs concerted efforts from heath service managers,
pathologists and IT specialist, along with defining standards for image quality, image formats and the underlying
hospital infrastructure required for routine digital microscopy.
A teleconsultation setting is also useful for re-evaluating diagnosed cases for quality assurance purposes [37]. In an
exercise involving 329 consecutive cases, high concordance (>91,8%) with the original pathology report was achieved
when retested using digital microscopy. Major discrepancies were found in only 1,5% of cases with none of them
requiring altered patient management. External quality assurance programs such as those run by the UK National
External Quality Assurance Scheme (UK NEQAS) and the similar Hungarian organization (QualiCont; see at
www.pathonet.com) have used digital slides for testing interpretational skills in hematology with good concordance
with the result of glass slide surveys [38].
Furthermore, digital slides can support recent efforts of system integration in histopathology to serve the
improvement of standards, traceability and reproducibility i.e. quality assurance of diagnostic procedures. System
integration involves network linking of procedures and sample data of all units within histopathology laboratories for
tracing of diagnostic samples/slides from entry to reporting and monitoring indicators of quality. Digital microscopy
can also assist virtual tumor banks such as the TuBaFrost (The European human frozen tumor tissue bank) offering a
digital slide catalogue of tumor samples stored at distant laboratories [39].
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
©FORMATEX 2010
971
______________________________________________
7. Automated image analysis – object quantification and pattern recognition
Empirical analysis of signals of molecular morphology techniques is tedious and prone to intra- and interobserver
variability, which result in low reproducibility and limited statistical confidence [15]. Recognition by the human eye of
microscopic objects such as cells, cell compartments (nucleus, cytoplasm and membrane) is based on distinguishing
groups of similar quality (color, intensity and size) units and their borders where these qualities markedly change.
However this recognition can be biased (Fig. 6). Digital slides are built up of discrete pixels of calibrated such qualities,
which are accurately registered by the computer. These can be used for image segmentation-based automated
recognition and quantification of cells, cell compartments (e.g. nucleus), tissue structures (e.g. glands), and their related
specific molecular signals.
Fig. 6 Recognition of color and
intensity by the naked eye can
be biased. As an example, the
same single colored vertical bar
appears homogenous against a
white background (left) but it
looks inhomogeneous against
changing gradient of colors
(right). After Conway et al,
2008.
In case of biomarker detection the signal intensity is proportional with the concentration of the molecule detected,
therefore, digital images can be used for automated biomarker quantification. Several integrated scanning and image
analysis systems are available for automated examination of immunohistochemistry and FISH signals [20]. In
combination with tissue microarrays (TMAs, see later) these can be used for high-throughput automated screening of
biomarker expression [40]. However, due to the biochemical complexity of cells and difficulties of standardizing cell
and tissue processing, calibrated absolute controls for tissue proteins are not yet available [8]. Therefore, only relative
(semi-quantitative) protein concentrations can be determined, which are still appropriate for doing comparative studies,
e.g. correlating biomarker levels with clinical behavior (tumor grade, survival etc…) [15]. Since FISH detects numerical
aberrations of chromosomes and genes its results can be quantified with high precision using multilayer scanning digital
microscopy in fluorescence mode (see above).
Stains are usually combined therefore, the pixels of a color range specifying a staining in cell/tissue structures must
be separated from other colors which also contribute to the image. An efficient brightfield unmixing (color-
deconvolution) algorithm can separate the contribution of up to three stains to the final color of an object (Fig. 7A) [41].
It can produce intensity maps of single target stains represented as monochrome images, which can be used to
determine surface area, overall absorption and density ranges in the specified objects. In practice, these can be e.g. the
brown signal of diaminobenzidine (DAB) in immunohistochemistry and the blue signal of hematoxylin nuclear
counterstaining, or different chromogenic signals of multiple immunolabelling. More precise optical spectra of pixels
and color separation can be acquired by using a spectral imaging camera that creates a stack of images at multiple
wavelengths [42]. The spatially resolved spectral information can substantially improve the linearity of quantification
and thus the utility of immunohistochemistry [8].
Image analysis systems offered for general purposes are mainly used in research and marker validation for
classifying cell types based on their size, shape, color and density. Specific algorithms for cell membrane
immunoreactions, mainly of the HER2 protein detection (Fig. 7B), and for nuclear signal detection, especially of ER
(estrogen receptor) (Fig. 7C) and PgR (progesterone receptor), are also available for the automated classification of
breast cancers and predicting their potential response to immuno- or hormonetherapy respectively. They also allow
post-processing analysis of the results by generating scatter plots (Fig. 7D), graphs and galleries and arrange data in an
XLS format which is well suited for statistical analysis [11]. Automated semi-quantitative analysis of HER2 protein
expression in breast cancer can improve test accuracy compared to manual scoring [43], which is recognized with a
reimbursement code in the USA by the American Medical Association. The increasing number of gene/chromosome
anomalies detected with FISH can also be analyzed using separate algorithms dedicated for the particular type of
abnormality e.g. amplification, translocation etc… Both bright-field and fluorescence algorithms are capable of sorting
the detected objects into galleries according to their distinct qualities, e.g. size, intensity and number of signals within
cells and re-localizing of individual objects from the gallery into the digital slide (see Fig. 4F).
Complex cytometric parameters allow automated differentiation of higher hierarchy structures such as normal and
diseased gastric glands based on linking cell nuclei into a cell web which demarcates the shape and size of glands for
automated classification [17, 44].
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
972
©FORMATEX 2010
______________________________________________
Fig. 7 A) The color deconvolution algorithm allows an accurate separation of target stains from each other such as the DAB
(diaminobenzidine) signal of HER2 immunostaining from the hematoxylin counterstain in breast cancer cells. B) Automated analysis
of cell membrane reactions classifies HER2 positive cells into 1+ (yellow), 2+ (orange) and 3+ (red) categories based on their
reaction intensity and membrane continuity. The frequency distribution of such categories will decide the whole case, which is e.g.
3+ when >30% of cells are 3+. C) Similar automated analysis can classify cells based on their nuclear immunoreactions, such as that
of estrogen receptor (ER) in breast cancer. D) Detected objects within a selected area on C can be further analyzed and gated using
scatter plots, by correlating for instance, color intensity (Bint) and object size (Area).
8. Supporting high throughput research using tissue microarrays (TMA)
Compared to the normal tissues disease processes are featured by altered morphology and molecular profile.
Translational research efforts focus specific attention on revealing molecular associations of cancer development and
progression and how these changes relate to the biological behavior (prognosis) and the potential response of tumors to
molecular target therapy (prediction) [45]. Large amount of such data have been generated at the DNA and mRNA
levels with high throughput comparative genomic hybridization (CGH) and expression chip array techniques, which
require validation at the gene and protein level in large numbers of relevant cases [3]. Tissue microarray (TMA)
technique allows the simultaneous staining and analysis of over 500 tissue cores arranged on the same glass slide and
thus thousands of tissue samples can be tested efficiently for a biomarker when slides from a few of such array blocks
are used [46]. With utilizing TMAs one can save substantial time, reagent, workforce and prestigious tissue, and the
technique allows high level of method standardization since samples on the same glass slide are handled under the very
same conditions. TMAs have been extensively used for high throughput biomarker screening and validation studies
both in basic and preclinical cancer research and in toxicology testing [47, 48, 49]. TMAs are also applied by the
“Human protein atlas” (http://www.proteinatlas.org) for systemic screening of the expression pattern and localization of
human proteins in normal and tumor tissues and the results are freely accessible through digital microscopy [50].
In TMA studies enormous number of staining, clinicopathological and image data have to be linked and analyzed
safely. This is extremely tedious and almost impossible to do with conventional optical microscopy which carries the
inherent risk of loosing track while also lacking the potential for standardized scoring and result validation. Digital
TMA slides can be integrated into common databases including all sample data, which then are accessed with software
tools dedicated to support fluency and consistency of TMA studies [51]. Most web-based open-access TMA managers
such as TMAD [52] and TAMME [53]; or commercial TMA systems such as the TMAx [50] use digital still images to
build large TMA data repositories. Low power still images, however, may hinder access to critical sample details.
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
©FORMATEX 2010
973
______________________________________________
Advanced TMA systems such as the Pannoramic TMA Module (previously called Mirax) [54], TMA LabII [55], Ariol
[56] and TMAscore [57] utilize proprietary digital slide formats and offer automated image analysis and quantification
options.
In contrast with most TMA managers using single slide at a time, the Pannoramic system is project-based and
supports the complete TMA workflow from design through creation of TMAs to the analysis of staining results (Fig. 8)
[51]. Within one project it can simultaneously manage consecutive slides cut from series of TMA blocks which are
stained for different biomarkers either using bright-field or fluorescence techniques. Furthermore, the system allows
setting up individualized scoring scheme for each marker, which can be tested by several assessors independently, and
to consolidate the scores eventually. The most accurate quantitation can be done with the automated image analysis
options [40] available for both bright-field and fluorescence signals (see above), where digital slides facilitate batch
processing of many areas of interest as it is shown in a project counting osteoclastic giant cells in relation to epidermal
growth factor receptor (EGFR) expression in giant cell tumor of bone [58].
Fig. 8 Basic workflow of the project-based Pannoramic TMA system. The TMA project starts with case selection and pooling of
relevant sample data into an excel database. Tissue cores from donor blocks are collected into recipient TMA blocks using a
computer driven TMA builder instrument (1), which also links sample data to their positions in the TMA blocks. Following cutting
and staining (2) TMA slides are digitalized (3) for scoring and image analysis (4) with the TMA Module software. Excel data are
input both for TMA building and analysis. Final results are arranged in the database for straightforward statistical or cluster analysis
(5).
At the start of a Pannoramic TMA project a computer driven instrument is used to design TMA layout, drill recipient
blocks and punch and insert tissue cores into the TMAs from the donor blocks (Fig. 8) [51]. Sample data in XLS format
are imported and linked to recipient block positions automatically at insertion, which prevent confusing the sample
order. Core areas of interest can be pre-selected and labeled on digital H&E slides, which will then be overlaid by the
software on top of the donor block’s image to guarantee accurate sampling. The TMA module software using digital
TMA slides allows semi-automated core finding, fluent scoring and rearrangement of stained spots independently from
their original physical location by selecting any modifier (e.g. diagnosis, score etc…) included in the XLS database, for
re-evaluation and final validation of the results. All data generated in a project are summarized in a common output file
which can be easily consolidated for statistics using any commercial software package.
The principle of TMA of analyzing only selected areas of a slide can also be adapted to digital whole slide studies.
For this, consecutive whole slides can be virtually fragmented after aligning them on top of each other and placing grids
on the areas of interest. Then the same areas stained for different biomarkers can be selected and analyzed at high power
side-by-side using any of the options available for TMAs including image quantification that improves the reliability of
comparative screening studies.
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
974
©FORMATEX 2010
______________________________________________
9. Concluding remarks
The tradition of centuries of using optical microscopy for studying tissue and cell morphology has been changing most
recently. Digital microscopy takes advantage of the multi-functionality of computer technology for rapidly accessing
and analyzing information embedded in stained slides, building user-friendly virtual slide and data repositories, and
allowing remote access to comprehensive databases including digital slides. In histopathology, morphology has been
recently assigned increasing amount of tests generating molecular information in situ, which must be analyzed with
high accuracy since it may directly determine patients’ chances for specific treatment and recovery. Also, the
accumulating knowledge on disease processes has turned pathology into a highly sub-specialized discipline. Digital
microscopy offers strong support in the huge challenge of coping with the growing workload and sub-specialization at
the increasing expectations from clinicians and patients by roughly constant number of staff. Digital microscopy can
boost the efficiency of histopathology both in diagnostic and research applications. These include graduate- and
postgraduate teaching, telepathology and -consultation, proficiency testing, intra- and interlaboratory quality assurance,
high throughput biomarker screening, and automated image analysis and quantitation for unbiased interpretation of
staining results. However, for the general acceptance of this new technology substantial efforts are still required in
standardizing digital slide file formats and informatics (DICOM) [1], hospital database systems and workflow (IHE)
[2], optimizing workstations and interfaces [29], further accelerating scanning and network speed [32], training of
professionals for digital microscopy [59] and addressing medico-legal aspects [30]. Most of these issues can be resolved
in the next 5-10 years, which will permit digital microscopy to make substantial contribution to modernizing
histopathology and improving patient care.
Acknowledgements The authors are indebted to Edit Parsch for excellent technical assistance and to Dr Peter Gombas, Laszlo
Krecsak, Laszlo Gerely and Daniel Szabo for useful consultation.
References
[1] Kayser K, Molnar B, Weinstein RS. Virtual slides technology. In: Kayser K, Molnar B, Weinstein RS eds. Virtual microscopy:
Fundamentals, applications, perspectives of electronic tissue-based diagnosis. Berlin, VSV Publ.; 2006:103-123.
[2] Weinstein RS, Graham AR, Richter LC, Barker GP, Krupinski EA, Lopez AM, Erps KA, Bhattacharyya AK, Yagi Y,
Gilbertson JR. Overview of telepathology, virtual microscopy, and whole slide imaging: prospects for the future. Hum Pathol.
2009; 40:1057-1069.
[3] Ma WW, Adjei AA. Novel agents on the horizon for cancer therapy. CA Cancer J Clin. 2009; 59:111-137.
[4] Zwick E, Bange J, Ullrich A. Receptor tyrosine kinase signalling as a target for cancer intervention strategies. Endocr Relat
Cancer. 2001; 8:161-173.
[5] Pfeifer JD, Hill DA, O'Sullivan MJ, Dehner LP. Diagnostic gold standard for soft tissue tumours: morphology or molecular
genetics? Histopathology. 2000; 37:485-500.
[6] Vorsanova SG, Yurov YB, Iourov IY. Human interphase chromosomes: a review of available molecular cytogenetic
technologies. Mol Cytogenet. 2010; 3:1. doi:10.1186/1755-8166-3-1
[7] Bodey B. The significance of immunohistochemistry in the diagnosis and therapy of neoplasms. Expert Opin Biol Ther. 2002;
2:371-393.
[8] Taylor CR, Levenson RM. Quantification of immunohistochemistry - issues concerning methods, utility and semiquantitative
assessment II. Histopathology. 2006; 49:411-424.
[9] Gabril MY, Yousef GM. Informatics for practicing anatomical pathologists: marking a new era in pathology practice. Mod
Pathol. 2010; 23:349-358.
[10] Kang HP, Sirintrapun SJ, Nestler RJ, Parwani AV. Experience with voice recognition in surgical pathology at a large academic
multi-institutional center. Am J Clin Pathol. 2010; 133:156-159.
[11] Krenacs T, Zsakovics I, Diczhazi C, Ficsor L, Varga VS, Molnar B. The potential of digital microscopy in breast pathology.
Pathol Oncol Res. 2009; 15:55-58.
[12] Rocha R, Vassallo J, Soares F, Miller K, Gobbi H. Digital slides: present status of a tool for consultation, teaching, and quality
control in pathology. Pathol Res Pract. 2009; 205:735-741.
[13] May M. A better lens on disease. Computerized pathology slides may help doctors make faster and more accurate diagnoses.
Scientific American. 2010; May:74-77.
[14] Rojo MG, García GB, Mateos CP, García JG, Vicente MC. Critical comparison of 31 commercially available digital slide
systems in pathology. Int J Surg Pathol. 2006; 14:285-305.
[15] Mulrane L, Rexhepaj E, Penney S, Callanan JJ, Gallagher WM. Automated image analysis in histopathology: a valuable tool in
medical diagnostics. Expert Rev Mol Diagn. 2008; 8:707-725.
[16] Weinstein RS, Descour MR, Liang C, Barker G, Scott KM, Richter L, Krupinski EA, Bhattacharyya AK, Davis JR, Graham
AR, Rennels M, Russum WC, Goodall JF, Zhou P, Olszak AG, Williams BH, Wyant JC, Bartels PH. An array microscope for
ultrarapid virtual slide processing and telepathology. Design, fabrication, and validation study. Hum Pathol. 2004; 35:1303-
1314.
[17] Ficsor L, Varga V, Berczi L, Miheller P, Tagscherer A, Wu ML, Tulassay Z, Molnar B. Automated virtual microscopy of
gastric biopsies. Cytometry B Clin Cytom. 2006; 70:423-431.
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
©FORMATEX 2010
975
______________________________________________
[18] Wu ML, Varga VS, Kamaras V, Ficsor L, Tagscherer A, Tulassay Z, Molnar B. Three-dimensional virtual microscopy of
colorectal biopsies. Arch Pathol Lab Med. 2005; 129:507-510.
[19] Herman B. Fluorescence microscopy. Curr Protoc Cell Biol. 2002; Chapter 4:Unit 4.2. DOI: 10.1002/0471143030.cb0402s13
[20] Rojo MG, Bueno G, Slodkowska J. Review of imaging solutions for integrated quantitative immunohistochemistry in the
Pathology daily practice. Folia Histochem Cytobiol. 2009; 47:349-354.
[21] Elliott K, Hamilton PW, Maxwell P. Fluorescence (FISH) and chromogenic (CISH) in situ hybridisation in prostate carcinoma
cell lines: comparison and use of virtual microscopy. Br J Biomed Sci. 2008; 65:167-171.
[22] Tanas MR, Goldblum JR. Fluorescence in situ hybridization in the diagnosis of soft tissue neoplasms: a review. Adv Anat
Pathol. 2009; 16:383-391.
[23] Varga VS, Ficsor L, Kamarás V, Jónás V, Virág T, Tulassay Z, Molnár B. Automated multichannel fluorescent whole slide
imaging and its application for cytometry. Cytometry A. 2009; 75:1020-1030.
[24] Dee FR. Virtual microscopy in pathology education. Hum Pathol. 2009; 40:1112-1121.
[25] Cserneky M, Szende B, Fonyad L, Krenacst T: Telepathology in Hungary. In: Telepathology. Kumar S and Dunn BE eds.
Springer-Verlag, Berlin, 2009; pp127-146.
[26] Bruch LA, De Young BR, Kreiter CD, Haugen TH, Leaven TC, Dee FR. Competency assessment of residents in surgical
pathology using virtual microscopy. Hum Pathol. 2009; 40:1122-1128.
[27] Dee FR, Donnelly A, Radio S, Leaven T, Zaleski MS, Kreiter C. Utility of 2-D and 3-D virtual microscopy in cervical cytology
education and testing. Acta Cytol. 2007; 51:523-529.
[28] 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; 57:1288-1291.
[29] Krupinski EA. Virtual slide telepathology workstation-of-the-future: lessons learned from teleradiology. Semin Diagn Pathol.
2009; 26:194-205.
[30] Leung ST, Kaplan KJ. Medicolegal aspects of telepathology. Hum Pathol. 2009; 40:1137-1142.
[31] Lundin M, Szymas J, Linder E, Beck H, de Wilde P, van Krieken H, García Rojo M, Moreno I, Ariza A, Tuzlali S, Dervisoglu
S, Helin H, Lehto VP, Lundin J. A European network for virtual microscopy-design, implementation and evaluation of
performance. Virchows Arch. 2009; 454:421-429.
[32] Wienert S, Beil M, Saeger K, Hufnagl P, Schrader T. Integration and acceleration of virtual microscopy as the key to
successful implementation into the routine diagnostic process. Diagn Pathol. 2009; 4:3. DOI:10.1186/1746-1596-4-3.
[33] Evans AJ, Chetty R, Clarke BA, Croul S, Ghazarian DM, Kiehl TR, Ordonez BP, Ilaalagan S, Asa SL. Primary frozen section
diagnosis by robotic microscopy and virtual slide telepathology: the University Health Network experience. Semin Diagn
Pathol. 2009; 26:165-176.
[34] Molnar B, Berczi L, Diczhazy C, Tagscherer A, Varga SV, Szende B, Tulassay Z. Digital slide and virtual microscopy based
routine and telepathology evaluation of routine gastrointestinal biopsy specimens. J Clin Pathol. 2003; 56:433-438.
[35] Lopez AM, Graham AR, Barker GP, Richter LC, Krupinski EA, Lian F, Grasso LL, Miller A, Kreykes LN, Henderson JT,
Bhattacharyya AK, Weinstein RS. Virtual slide telepathology enables an innovative telehealth rapid breast care clinic. Hum
Pathol. 2009; 40:1082-1091.
[36] Schrader T, Hufnagl P, Schlake W, Dietel M. Study of efficiency of teleconsultation: the Telepathology Consultation Service
of the Professional Association of German Pathologists for the screening program of breast carcinoma (in German). Verh Dtsch
Ges Pathol. 2005; 89:211–218.
[37] Graham AR, Bhattacharyya AK, Scott KM, Lian F, Grasso LL, Richter LC, Carpenter JB, Chiang S, Henderson JT, Lopez
AM, Barker GP, Weinstein RS. Virtual slide telepathology for an academic teaching hospital surgical pathology quality
assurance program. Hum Pathol. 2009; 40:1129-1136.
[38] Burthem J, Brereton M, Ardern J, Hickman L, Seal L, Serrant A, Hutchinson CV, Wells E, McTaggart P, De la Salle B,
Parker-Williams J, Hyde K. The use of digital 'virtual slides' in the quality assessment of haematological morphology: results of
a pilot exercise involving UK NEQAS(H) participants. Br J Haematol. 2005; 130:293-296.
[39] Teodorovic I, Isabelle M, Carbone A, Passioukov A, Lejeune S, Jaminé D, Therasse P, Gloghini A, Dinjens WN, Lam KH,
Oomen MH, Spatz A, Ratcliffe C, Knox K, Mager R, Kerr D, Pezzella F, van Damme B, van de Vijver M, van Boven H,
Morente MM, Alonso S, Kerjaschki D, Pammer J, Lopez-Guerrero JA, Llombart Bosch A, van Veen EB, Oosterhuis JW,
Riegman PH. TuBaFrost 6: virtual microscopy in virtual tumour banking. Eur J Cancer.2006; 42:3110-3116.
[40] Conway C, Dobson L, O'Grady A, Kay E, Costello S, O'Shea D. Virtual microscopy as an enabler of automated/quantitative
assessment of protein expression in TMAs. Histochem Cell Biol. 2008; 130:447-463.
[41] Ruifrok AC, Katz RL, Johnston DA. Comparison of quantification of histochemical staining by hue-saturation-intensity (HSI)
transformation and color-deconvolution. Appl Immunohistochem Mol Morphol. 2003; 11:85-91.
[42] Levenson RM. Spectral imaging perspective on cytomics. Cytometry A. 2006; 69:592-600.
[43] Bloom K, Harrington D. Enhanced Accuracy of HER-2/neu immunohistochemical scoring using digital microscopy. Am J Clin
Pathol. 2004; 121:619-630.
[44] Ficsor L, Varga VS, Tagscherer A, Tulassay Z, Molnar B. Automated classification of inflammation in colon histological
sections based on digital microscopy and advanced image analysis. Cytometry A. 2008; 73:230-237.
[45] Dy GK, Adjei AA. Systemic cancer therapy: evolution over the last 60 years. Cancer. 2008; 113(7 Suppl):1857-1887.
[46] Kononen, J., Bubendorf, L., Kallioniemi, A., Barlund, M., Schraml, P., Leighton, S., Torhorst, J., Mihatsch, M.J., Sauter, G.,
Kallioniemi, O.P. Tissue microarrays for high through-put molecular profiling of tumor specimens. Nat. Med. 1998; 4:844-847.
[47] Whiteford CC, Bilke S, Greer BT, Chen Q, Braunschweig TA, Cenacchi N, Wei JS, Smith MA, Houghton P, Morton C,
Reynolds CP, Lock R, Gorlick R, Khanna C, Thiele CJ, Takikita M, Catchpoole D, Hewitt SM, Khan J. Credentialing
preclinical pediatric xenograft models using gene expression and tissue microarray analysis. Cancer Res. 2007; 67:32-40.
[48] Mulrane L, Rexhepaj E, Smart V, Callanan JJ, Orhan D, Eldem T, Mally A, Schroeder S, Meyer K, Wendt M, O'Shea D,
Gallagher WM. Creation of a digital slide and tissue microarray resource from a multi-institutional predictive toxicology study
in the rat: an initial report from the PredTox group. Exp Toxicol Pathol. 2008; 60:235-245.
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
976
©FORMATEX 2010
______________________________________________
[49] Potts SJ. Digital pathology in drug discovery and development: multisite integration. Drug Discov Today. 2009; 14:935-941.
[50] Hober, S. and Uhlen, M. Human protein atlas and the use of microarray technologies. Curr Opin Biotechnol. 2008; 19:30–35.
DOI:10.1016/j.copbio. 2007.11.006.
[51] Krenacs T, Ficsor L, Varga VS, Angeli V, Molnar B. Digital microscopy for boosting database integration and analysis in
TMA studies. Methods in Molecular Biology. (Ed: Simon R), Humana Press, Totowa, NJ (in press).
[52] Marinelli RJ, Montgomery K, Liu CL, Shah NH, Prapong W, Nitzberg M, Zachariah ZK, Sherlock GJ, Natkunam Y, West RB,
van de Rijn M, Brown PO, Ball CA. The Stanford Tissue Microarray Database. Nucleic Acids Res. 2008; 36:D871-817.
[53] Thallinger GG, Baumgartner K, Pirklbauer M, Uray M, Pauritsch E, Mehes G, Buck CR, Zatloukal K, Trajanoski Z. TAMEE:
data management and analysis for tissue microarrays. BMC Bioinformatics. 2007; 8:81. DOI:10.1186/1471-2105-8-81.
[54] Stelkovics E, Korom I, Marczinovits I, Molnar J, Rasky K, Raso E, Ficsor E, Molnar B, Kopper L, Krenacs T. Collagen
XVII/BP180 protein expression in squamous cell carcinoma of the skin detected with novel monoclonal antibodies in archived
tissues using tissue microarrays and digital microscopy. Appl Immunohistochem Mol Morphol. 2008; 16:433–441.
[55] Anderson WF, Luo S, Chatterjee N, Rosenberg PS, Matsuno RK, Goodman MT, Hernandez BY, Reichman M, Dolled-Filhart
MP, O'Regan RM, Garcia-Closas M, Perou CM, Jatoi I, Cartun RW, Sherman ME. Human epidermal growth factor receptor-2
and estrogen receptor expression, a demonstration project using the residual tissue repository of the Surveillance,
Epidemiology, and End Results (SEER) program. Breast Cancer Res Treat. 2009; 113:189-196.
[56] Turbin DA, Leung S, Cheang MC, Kennecke HA, Montgomery KD, McKinney S, Treaba DO, Boyd N, Goldstein, LC, Badve
S, Gown AM, van de Rijn M, Nielsen TO, Gilks CB, Huntsman DG. Automated quantitative analysis of estrogen receptor
expression in breast carcinoma does not differ from expert pathologist scoring: a tissue microarray study of 3,484 cases. Breast
Cancer Res Treat. 2008;110:417-426.
[57] Rubin MA, Dunn R, Strawderman M, Pienta KJ. Tissue microarray sampling strategy for prostate cancer biomarker analysis.
Am J Surg Pathol. 2002; 26:312-319.
[58] Balla P, Moskovszky L, Sapi Z, Forsyth R, Knowles H, Athanasou NA, Szendroi M, Kopper L, Rajnai H, Pinter F, Petak I,
Benassi MS, Picci P, Conti A, Krenacs T. Epidermal growth factor receptor signalling contributes to osteoblastic stromal cell
proliferation, osteoclastogenesis and disease progression in giant cell tumour of bone. Histopathology 2010; (accepted for
publication).
[59] Evered A, Dudding N. Accuracy and perceptions of virtual microscopy compared with glass slide microscopy in cervical
cytology. Cytopathology. 2010 May 17. [Epub ahead of print]
Microscopy: Science, Technology, Applications and Education
A. Méndez-Vilas and J. Díaz (Eds.)
©FORMATEX 2010
977
______________________________________________
